|
22545
|
972
|
37
|
2026-05-12T07:13:03.097163+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778569983097_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.09652778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.19861111,"height":0.10944445},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.09861111,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.20173611,"height":0.10944445},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"bounds":{"left":0.16527778,"top":0.19944444,"width":0.017708333,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"bounds":{"left":0.18715277,"top":0.19777778,"width":0.11215278,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"bounds":{"left":0.123958334,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"bounds":{"left":0.14618056,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"bounds":{"left":0.16840278,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-2650477813380124461
|
-2831770885395858542
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22546
|
972
|
38
|
2026-05-12T07:13:06.098602+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778569986098_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-3188013104974007810
|
-3011479387632878973
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar...
|
22545
|
NULL
|
NULL
|
NULL
|
|
22547
|
973
|
21
|
2026-05-12T07:13:06.325373+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778569986325_m2.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.042220745,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5171588,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.5442937,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
8009247334277766736
|
-2831911623152649326
|
idle
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions...
|
22537
|
NULL
|
NULL
|
NULL
|
|
22548
|
972
|
39
|
2026-05-12T07:13:12.180402+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778569992180_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
8926275643889037655
|
-2831770884858987630
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22549
|
972
|
40
|
2026-05-12T07:13:15.177134+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778569995177_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-8243821770060575709
|
-2444025834583934333
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)...
|
22548
|
NULL
|
NULL
|
NULL
|
|
22550
|
972
|
41
|
2026-05-12T07:13:18.226293+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778569998226_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
8432176313194120725
|
-2579134922916660606
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22551
|
973
|
22
|
2026-05-12T07:13:36.776553+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570016776_m2.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.042220745,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5171588,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.5442937,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-1285790672158704186
|
-2831913822176429422
|
idle
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22552
|
972
|
42
|
2026-05-12T07:13:39.371574+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570019371_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.09652778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.19861111,"height":0.10944445},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.09861111,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.20173611,"height":0.10944445},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"bounds":{"left":0.16527778,"top":0.19944444,"width":0.017708333,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"bounds":{"left":0.18715277,"top":0.19777778,"width":0.11215278,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"bounds":{"left":0.123958334,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"bounds":{"left":0.14618056,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"bounds":{"left":0.16840278,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"bounds":{"left":0.190625,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"bounds":{"left":0.21284722,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.15729167,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"bounds":{"left":0.18784723,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
1119763493609424136
|
-2831770885395858542
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit...
|
22550
|
NULL
|
NULL
|
NULL
|
|
22553
|
972
|
43
|
2026-05-12T07:13:43.570929+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570023570_m1.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
iTerm2ShellEditViewSessionScriptsProfilesWindowHel iTerm2ShellEditViewSessionScriptsProfilesWindowHelpSupport Daily - in 4 h 47 mlmeet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.comNikolay Yankov (Presenting, annotating)ChromeFileViewHistoryTab1 SeJirWindowHelpX Prinry4 AtAt:*Jin ClС м8• Tue 12 May 10:138 мс8 Jirlhttps://jiminny.atlassian.net/jira/software/c/projects/JY/boards/37© AWS|X) SSH• Home | SalesforcePlatform TeamQ Search boardEpic vType vQuick filters vREADY FOR DEVINDEV 3CODE REVIEW 2BLOCKEDUpgrade to PHP 8.5PHP 8.5 UPGRADEIn Dev3 П •••=AJ Panorama forCall Scoring in ODAUTOMATED AI SCORINGCode Review2.5 1 =E JY-18091|[ JY-20361|SAI Review - Q1 -Summary/Actionitems/Key PointsGROWTH - MAINTAIN OU...In Dev[HubSpot] OptimiseCRM rematching ondelete hubspot…PLATFORM STABILITYCode Review2 @ •=.4 h =• *[ JY-20566|X JY-20725[POC)Jiminny MCPConnectorJIMINNY MCP CONNECTORIn Progress10 1 [PASSWORD_DOTS]=@JY-20625A[л|+Datadog* Claude|3 CircleciSentry0 All BookmarksComplete sprintGroup: Queries|QA 1PO ACCEPTANCEDEPLOY 10Smart Instant|Nudge Pre-filteringCOST-EFFECTIVE AND FA...In QA15 h =AI Reports > Emptypage design andpromotionAJ REPORTSDeployed1 П [PASSWORD_DOTS]=4 JY-20495• JY-20372Sync opportunitieswithout a localowner (user_id is...PLATFORM STABILITYDeployed5 1=9 JY-20352meet.google.comNikolay Yankov (You, present...Grok via AzureMAINTENANCEDeployed0Ф000-A JY-20726Allow users todelete SS andPanorama prompt...• Tue 12 May 10:13:438Nikolay IvanovNikolay Yankov3 othersStefka Stoyanova10:13 AM | Daily - PlatformLukas Kovalik27:5011!...
|
NULL
|
6955031722999763509
|
NULL
|
click
|
ocr
|
NULL
|
iTerm2ShellEditViewSessionScriptsProfilesWindowHel iTerm2ShellEditViewSessionScriptsProfilesWindowHelpSupport Daily - in 4 h 47 mlmeet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.comNikolay Yankov (Presenting, annotating)ChromeFileViewHistoryTab1 SeJirWindowHelpX Prinry4 AtAt:*Jin ClС м8• Tue 12 May 10:138 мс8 Jirlhttps://jiminny.atlassian.net/jira/software/c/projects/JY/boards/37© AWS|X) SSH• Home | SalesforcePlatform TeamQ Search boardEpic vType vQuick filters vREADY FOR DEVINDEV 3CODE REVIEW 2BLOCKEDUpgrade to PHP 8.5PHP 8.5 UPGRADEIn Dev3 П •••=AJ Panorama forCall Scoring in ODAUTOMATED AI SCORINGCode Review2.5 1 =E JY-18091|[ JY-20361|SAI Review - Q1 -Summary/Actionitems/Key PointsGROWTH - MAINTAIN OU...In Dev[HubSpot] OptimiseCRM rematching ondelete hubspot…PLATFORM STABILITYCode Review2 @ •=.4 h =• *[ JY-20566|X JY-20725[POC)Jiminny MCPConnectorJIMINNY MCP CONNECTORIn Progress10 1 [PASSWORD_DOTS]=@JY-20625A[л|+Datadog* Claude|3 CircleciSentry0 All BookmarksComplete sprintGroup: Queries|QA 1PO ACCEPTANCEDEPLOY 10Smart Instant|Nudge Pre-filteringCOST-EFFECTIVE AND FA...In QA15 h =AI Reports > Emptypage design andpromotionAJ REPORTSDeployed1 П [PASSWORD_DOTS]=4 JY-20495• JY-20372Sync opportunitieswithout a localowner (user_id is...PLATFORM STABILITYDeployed5 1=9 JY-20352meet.google.comNikolay Yankov (You, present...Grok via AzureMAINTENANCEDeployed0Ф000-A JY-20726Allow users todelete SS andPanorama prompt...• Tue 12 May 10:13:438Nikolay IvanovNikolay Yankov3 othersStefka Stoyanova10:13 AM | Daily - PlatformLukas Kovalik27:5011!...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22554
|
972
|
44
|
2026-05-12T07:13:45.548643+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570025548_m1.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
Nikolay Yankov (Presenting, annotating)
Nikolay Yankov (Presenting, annotating)
People
8
Take notes with Gemini
Take notes with Gemini
Gemini
Gemini
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Unpin Nikolay Yankov's presentation from your main screen
You can't unmute someone else's presentation
More options for Nikolay Yankov
Zoom in
Open in new window
Enter Full Screen
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Pin Nikolay Ivanov to your main screen
Mute Nikolay Ivanov's microphone
More options for Nikolay Nikolov
Nikolay Ivanov
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Meet - Daily - Platform","depth":4,"bounds":{"left":0.0,"top":0.072222225,"width":0.033680554,"height":0.045555554},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.0013888889,"top":0.072222225,"width":0.010416667,"height":0.016666668},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.005902778,"top":0.12,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.7977778,"width":0.033680554,"height":0.043333333},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.8411111,"width":0.033680554,"height":0.038333334},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.0,"top":0.8794444,"width":0.033680554,"height":0.03888889},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.0,"top":0.91833335,"width":0.033680554,"height":0.038333334},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.95666665,"width":0.033680554,"height":0.043333333},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"Nikolay Yankov (Presenting, annotating)","depth":12,"bounds":{"left":0.07534722,"top":0.101111114,"width":0.17777778,"height":0.022222223},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Nikolay Yankov (Presenting, annotating)","depth":13,"bounds":{"left":0.07534722,"top":0.10222222,"width":0.17777778,"height":0.020555556},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"People","depth":15,"bounds":{"left":0.88680553,"top":0.08944444,"width":0.04097222,"height":0.04},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"8","depth":22,"bounds":{"left":0.9145833,"top":0.101111114,"width":0.0048611113,"height":0.017222222},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Take notes with Gemini","depth":14,"bounds":{"left":0.93333334,"top":0.08944444,"width":0.025,"height":0.04},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Take notes with Gemini","depth":17,"bounds":{"left":0.9361111,"top":0.101111114,"width":0.06388891,"height":0.017222222},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini","depth":22,"bounds":{"left":0.96666664,"top":0.101111114,"width":0.028125,"height":0.017222222},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Gemini","depth":21,"bounds":{"left":0.96458334,"top":0.090555556,"width":0.023611112,"height":0.037777778},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Pop out this video More screens are more fun. Play this video while you do other things.","depth":15,"bounds":{"left":0.5798611,"top":0.62833333,"width":0.14652778,"height":0.08888889},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pop out this video","depth":17,"bounds":{"left":0.7239583,"top":0.6427778,"width":0.08090278,"height":0.018888889},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"More screens are more fun. Play this video while you do other things.","depth":16,"bounds":{"left":0.7017361,"top":0.6388889,"width":0.11076389,"height":0.05666667},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Unpin Nikolay Yankov's presentation from your main screen","depth":13,"bounds":{"left":0.346875,"top":0.5083333,"width":0.027777778,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"You can't unmute someone else's presentation","depth":13,"bounds":{"left":0.37465277,"top":0.5061111,"width":0.030555556,"height":0.04888889},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options for Nikolay Yankov","depth":13,"bounds":{"left":0.40520832,"top":0.5083333,"width":0.027777778,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Zoom in","depth":13,"bounds":{"left":0.6315972,"top":0.83111113,"width":0.027777778,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open in new window","depth":13,"bounds":{"left":0.6649306,"top":0.83111113,"width":0.027777778,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Enter Full Screen","depth":13,"bounds":{"left":0.6982639,"top":0.83111113,"width":0.027777778,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Pop out this video More screens are more fun. Play this video while you do other things.","depth":15,"bounds":{"left":0.7861111,"top":0.27611113,"width":0.14652778,"height":0.07722222},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pop out this video","depth":17,"bounds":{"left":0.9302083,"top":0.2911111,"width":0.069791675,"height":0.017777778},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"More screens are more fun. Play this video while you do other things.","depth":16,"bounds":{"left":0.9079861,"top":0.28666666,"width":0.092013896,"height":0.045},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Pin Nikolay Ivanov to your main screen","depth":13,"bounds":{"left":0.76180553,"top":0.25111112,"width":0.027777778,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Mute Nikolay Ivanov's microphone","depth":13,"bounds":{"left":0.7895833,"top":0.2488889,"width":0.030555556,"height":0.04888889},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options for Nikolay Nikolov","depth":13,"bounds":{"left":0.8201389,"top":0.25111112,"width":0.027777778,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Nikolay Ivanov","depth":17,"bounds":{"left":0.75451386,"top":0.36277777,"width":0.07395833,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Pop out this video More screens are more fun. Play this video while you do other things.","depth":15,"bounds":{"left":0.91180557,"top":0.27611113,"width":0.08819443,"height":0.07722222},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pop out this video","depth":17,"bounds":{"left":1.0,"top":0.2911111,"width":-0.05590272,"height":0.017777778},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"More screens are more fun. Play this video while you do other things.","depth":16,"bounds":{"left":1.0,"top":0.28666666,"width":-0.03368056,"height":0.045},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-5668986924033736191
|
-6425962119525926092
|
visual_change
|
accessibility
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
Nikolay Yankov (Presenting, annotating)
Nikolay Yankov (Presenting, annotating)
People
8
Take notes with Gemini
Take notes with Gemini
Gemini
Gemini
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Unpin Nikolay Yankov's presentation from your main screen
You can't unmute someone else's presentation
More options for Nikolay Yankov
Zoom in
Open in new window
Enter Full Screen
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Pin Nikolay Ivanov to your main screen
Mute Nikolay Ivanov's microphone
More options for Nikolay Nikolov
Nikolay Ivanov
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things....
|
22553
|
NULL
|
NULL
|
NULL
|
|
22555
|
973
|
23
|
2026-05-12T07:13:43.545249+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570023545_m2.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
rireroxcalMISTOMbookmarksProtllesJOOISWindowmelprT rireroxcalMISTOMbookmarksProtllesJOOISWindowmelprTavsco.s?9 JY-20725-handle-HS-search-raProleteyC) AutomatedReportGenerated.onp© PlaybackController.php xC) SubscrintionControfinal class PlaybackController extends FrontenoC) TeamA AutomationC) TeamA ConteytConФ TeamController.phpc) ToaminciahtcContr© TranscriptionContrc(C) TranclationControllc usercontroller.pnp© VocabularyControll>@ Auth-customerapl›J Internav D Kioskeams© ActivityController.pAutomatedReportsic) DashboardControllec) ImpersonationContC) MediaPinelinecont0 OraanizationsContrC PartnersController.C) ProfileController.ohC)SearchController.of>• Settings>• Telephonyv M Wehhook> M Huhsnot>D IntegrationAppSub:© ActivityProviderCor© ActivityTranscriptio© BaseController.php© CalendarController,© ReportController.phSoftphoneWebhool.AhstractController.ong@ CommentContextinter© ConferencesOptInOutcc) controller.phpExportController.phpc)FrontendController.oh() FrontendControllerTra@ GeocodinaController.n(C) HealthCheckControllei@ LiveCoachController.p(C) Missina TeamControlle(C) MobileController.ohvC) NotificationController.t© NotificationProviderCoYe) PlavbackControlier ohg DlavlistController nhn(C) PucherController nhng SlackController nhnpudLlc tunction snow Activlcy$data = Fractal::createQ->item(SacuLVIEynStnancfonmen-scetConcumen/Sucel->serzal1zewichnew Jsonserallzer->cOArrayo"$data['playbackData']['masterPlaylist'if (! isset(Sdata['playbackData']['tradSdatal'playbackData']['tracks'] =* Sendina 'plaubackVisited' event to* response to the user e.a.lnften thedeferdfnO => Sthis->nlanhatService->tr:user: Suserevent: 'nlavbackVisited'payload: ['activityId' => $activity->lactivitvlluidi => Sactiviti)->alwaysO:Ireturn $this->renderCO'playbackData' =>['activity' => Sdatal'playbackDa"tavorited' => sdatal'favoritei'subscribed' => $datal'subscril'VleW' => Svleworivate function getPreloadedPlavistlActil* athrows AuthorizationExcentionGET Inlavlict[activitv) m2u8 Inlavlictlpublic function playlist(Activity Sactivit•••New Tabe Jy 20820 es reindex stream mode4 [JY-20725] [HubSpot] Optimise Cl• JY-20725 add HS rate limit handlilPipelines - jiminny/app• Pull requests • jiminny/appWY-207731 User Pilot not receivinJY-20773 fix user pilot tracking ofiNY-20776 Automated report - seTypeError: League|Flysystem\FilesA TypeError: League\Flysystem\FilesPlatform Sprint 3 02 - Platform Te• JY-20625 | JY-20742 | MCP POC kG LLM pricing overview with the mo:G olanhat - Gooale Search- New TabolannatGoogle GemintI'm on page"<tabTitle>JY-20361: Addcall scores in Panorama bystelivan-</tao lites> with"‹selection>@@ -4,7 +4,7...show thinkingHere is a summary of the changes in theprovided code diff.Al Calll Scorind Additions• Pre-computed Scores: Added strictinstructions to treat the ai_call_scoreas ore-comouree oala. Ine system musInterorer ano summarize the nrovidodscores rather than inventing orrecalculating them• Context Breakdown: * Short Context:Now disolavs the Al Scorecard name andthe overall Al Score (the averaae of per-rule scores).• Full Context: Details theai call score object, breakingcown individual rules. Fach ruleincludes a title, prompt, a 1-5 score, abriet justitication, and up to threetimestamoed soeaker reterencesO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini oAlrcan make micnkec col daliale cnockt Yottr drivaeu wdomSummarize pageGoogleolannaluAl ModeShort videosReviewsSponsored resultPlanhat - Customer Success PlatformManage CS in one place. Gain visibility, reduce churn, automate key journeys. Visibility, predictionautomation—one CS platform to run it allAll-in-one CS PlatformUnify CS ops in one platform. Predict churn. spot expansion. and automate iournevs.Churn Prediction and PlaybooksScale vour CS team without the chaos—see more, know more, do more in one platform.Advance Al FeaturesRuilt in Al tor advance tacks Get more done taster with AlIMultiple IntearationsEasilv connect with vour tech stack Easilv connect with existina toolsAccurate ReportingEasily access data via dashboards Easily visualize customer dataHide sponsored result^Planhat.httns://ww.nlanhat.comPlanhatA customer platform where commercial teams work better. We build and deploy frontier technologies-intearatina data. Al and collaboration—that empower ..AhounA workspace that allows you to monitor and manage your ….CaoturacExolore Planhat's powerful features. Unify vour data …• suppon Dally • In 4h 4/ m100% Lz• lue 1z May 10:13.43NORRMALMSee nhotosPlanhat50 ***** 1 review ©Software comnanv in Stockholm Swedenl• Wehsite* Directions* ReviewsW SaveL ShareainsrodilineallllonanAddress: Sveavägen 17, 111 57 Stockholm, SwedenHours: Onen 24 hoursSuggest an edit • Own this business?Add missing informationAdd nlaces nhone numberFrom P anhar"Planhat is helping hundreds of modern technology companies worldwidecenter their business around their customers in order to maximizecuctomer cuccoce and cuctomor lifotime value "# Send to your phoneProfilesYoullubeLinkedinX aiwitter)...
|
NULL
|
-2328305088073304638
|
NULL
|
click
|
ocr
|
NULL
|
rireroxcalMISTOMbookmarksProtllesJOOISWindowmelprT rireroxcalMISTOMbookmarksProtllesJOOISWindowmelprTavsco.s?9 JY-20725-handle-HS-search-raProleteyC) AutomatedReportGenerated.onp© PlaybackController.php xC) SubscrintionControfinal class PlaybackController extends FrontenoC) TeamA AutomationC) TeamA ConteytConФ TeamController.phpc) ToaminciahtcContr© TranscriptionContrc(C) TranclationControllc usercontroller.pnp© VocabularyControll>@ Auth-customerapl›J Internav D Kioskeams© ActivityController.pAutomatedReportsic) DashboardControllec) ImpersonationContC) MediaPinelinecont0 OraanizationsContrC PartnersController.C) ProfileController.ohC)SearchController.of>• Settings>• Telephonyv M Wehhook> M Huhsnot>D IntegrationAppSub:© ActivityProviderCor© ActivityTranscriptio© BaseController.php© CalendarController,© ReportController.phSoftphoneWebhool.AhstractController.ong@ CommentContextinter© ConferencesOptInOutcc) controller.phpExportController.phpc)FrontendController.oh() FrontendControllerTra@ GeocodinaController.n(C) HealthCheckControllei@ LiveCoachController.p(C) Missina TeamControlle(C) MobileController.ohvC) NotificationController.t© NotificationProviderCoYe) PlavbackControlier ohg DlavlistController nhn(C) PucherController nhng SlackController nhnpudLlc tunction snow Activlcy$data = Fractal::createQ->item(SacuLVIEynStnancfonmen-scetConcumen/Sucel->serzal1zewichnew Jsonserallzer->cOArrayo"$data['playbackData']['masterPlaylist'if (! isset(Sdata['playbackData']['tradSdatal'playbackData']['tracks'] =* Sendina 'plaubackVisited' event to* response to the user e.a.lnften thedeferdfnO => Sthis->nlanhatService->tr:user: Suserevent: 'nlavbackVisited'payload: ['activityId' => $activity->lactivitvlluidi => Sactiviti)->alwaysO:Ireturn $this->renderCO'playbackData' =>['activity' => Sdatal'playbackDa"tavorited' => sdatal'favoritei'subscribed' => $datal'subscril'VleW' => Svleworivate function getPreloadedPlavistlActil* athrows AuthorizationExcentionGET Inlavlict[activitv) m2u8 Inlavlictlpublic function playlist(Activity Sactivit•••New Tabe Jy 20820 es reindex stream mode4 [JY-20725] [HubSpot] Optimise Cl• JY-20725 add HS rate limit handlilPipelines - jiminny/app• Pull requests • jiminny/appWY-207731 User Pilot not receivinJY-20773 fix user pilot tracking ofiNY-20776 Automated report - seTypeError: League|Flysystem\FilesA TypeError: League\Flysystem\FilesPlatform Sprint 3 02 - Platform Te• JY-20625 | JY-20742 | MCP POC kG LLM pricing overview with the mo:G olanhat - Gooale Search- New TabolannatGoogle GemintI'm on page"<tabTitle>JY-20361: Addcall scores in Panorama bystelivan-</tao lites> with"‹selection>@@ -4,7 +4,7...show thinkingHere is a summary of the changes in theprovided code diff.Al Calll Scorind Additions• Pre-computed Scores: Added strictinstructions to treat the ai_call_scoreas ore-comouree oala. Ine system musInterorer ano summarize the nrovidodscores rather than inventing orrecalculating them• Context Breakdown: * Short Context:Now disolavs the Al Scorecard name andthe overall Al Score (the averaae of per-rule scores).• Full Context: Details theai call score object, breakingcown individual rules. Fach ruleincludes a title, prompt, a 1-5 score, abriet justitication, and up to threetimestamoed soeaker reterencesO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini oAlrcan make micnkec col daliale cnockt Yottr drivaeu wdomSummarize pageGoogleolannaluAl ModeShort videosReviewsSponsored resultPlanhat - Customer Success PlatformManage CS in one place. Gain visibility, reduce churn, automate key journeys. Visibility, predictionautomation—one CS platform to run it allAll-in-one CS PlatformUnify CS ops in one platform. Predict churn. spot expansion. and automate iournevs.Churn Prediction and PlaybooksScale vour CS team without the chaos—see more, know more, do more in one platform.Advance Al FeaturesRuilt in Al tor advance tacks Get more done taster with AlIMultiple IntearationsEasilv connect with vour tech stack Easilv connect with existina toolsAccurate ReportingEasily access data via dashboards Easily visualize customer dataHide sponsored result^Planhat.httns://ww.nlanhat.comPlanhatA customer platform where commercial teams work better. We build and deploy frontier technologies-intearatina data. Al and collaboration—that empower ..AhounA workspace that allows you to monitor and manage your ….CaoturacExolore Planhat's powerful features. Unify vour data …• suppon Dally • In 4h 4/ m100% Lz• lue 1z May 10:13.43NORRMALMSee nhotosPlanhat50 ***** 1 review ©Software comnanv in Stockholm Swedenl• Wehsite* Directions* ReviewsW SaveL ShareainsrodilineallllonanAddress: Sveavägen 17, 111 57 Stockholm, SwedenHours: Onen 24 hoursSuggest an edit • Own this business?Add missing informationAdd nlaces nhone numberFrom P anhar"Planhat is helping hundreds of modern technology companies worldwidecenter their business around their customers in order to maximizecuctomer cuccoce and cuctomor lifotime value "# Send to your phoneProfilesYoullubeLinkedinX aiwitter)...
|
22551
|
NULL
|
NULL
|
NULL
|
|
22556
|
972
|
45
|
2026-05-12T07:13:47.228941+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570027228_m1.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Meet - Daily - Platform","depth":4,"bounds":{"left":0.0,"top":0.072222225,"width":0.033680554,"height":0.045555554},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.0013888889,"top":0.072222225,"width":0.010416667,"height":0.016666668},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.005902778,"top":0.12,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.7977778,"width":0.033680554,"height":0.043333333},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.8411111,"width":0.033680554,"height":0.038333334},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.0,"top":0.8794444,"width":0.033680554,"height":0.03888889},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.0,"top":0.91833335,"width":0.033680554,"height":0.038333334},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.95666665,"width":0.033680554,"height":0.043333333},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
6010633228612407689
|
1080959010570296966
|
click
|
hybrid
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
+FirefoxFileEditViewHistoryBookmarksProfilesToolsWindowHelp• 0(abl= Support Daily - in 4 h 47 m100% C78• Tue 12 May 10:13:46→meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.comNikolay Yankov (Presenting, annotating)ChromeFileViewHistory1 SeTabWindowHelpJirPrinry4 AtAt:*JinClС м3 мс8• Tue 12 May 10:138 Jielhttps://jiminny.atlassian.net/jira/software/c/projects/JY/boards/37X) SSHInonteoelealorePlatform TeamQ Search board|Epic vType vQuick filters vREADY FOR DEVIN DEV 3CODE REVIEW 2BLOCKEDUpgrade to PHP 8.5PHP 8.5 UPGRADEIn Dev3 П •••=AJ Panorama forCall Scoring in ODAUTOMATED AI SCORINGCode Review2.5 1 =E JY-18091|[ JY-20361|SAI Review - Q1 -Summary/Actionitems/Key PointsGROWTH - MAINTAIN OU...In Dev2 @ •=.[HubSpot] OptimiseCRM rematching ondelete hubspot…PLATFORM STABILITYCode Review• *Д JY-205664 h =X JY-20725[POC)Jiminny MCPConnectorJIMINNY MCP CONNECTORIn Progress10 1 [PASSWORD_DOTS]=• JY-20625|Datadog|MECIU» | C AIl Bookmarks|Complete sprintGroup: Queries|QA 1/PO ACCEPTANCEDEPLOY 10Smart Instant|Nudge Pre-filteringCOST-EFFECTIVE AND FA…..In QA15 h =AI Reports > Emptypage design andpromotionAJ REPORTSDeployed1 П [PASSWORD_DOTS]=4 JY-20495• JY-20372Sync opportunitieswithout a localowner (user_id is...PLATFORM STABILITYDeployed5 1=9 JY-20352meet.google.comNikolay Yankov (You, present...Grok via AzureMAINTENANCEDeployed0Ф000-A JY-20726Lowve calAllow users todelete SS andPanorama prompt...Leave callNikolay IvanovStefka Stoyanova=8Steliyan Georgiev3 others10:13 AM | Daily - PlatformLukas Kovalik27:53111...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22557
|
973
|
24
|
2026-05-12T07:13:47.775418+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570027775_m2.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
Nikolay Yankov (Presenting, annotating)
Nikolay Yankov (Presenting, annotating)
People
8
Take notes with Gemini
Take notes with Gemini
Gemini
Gemini
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Unpin Nikolay Yankov's presentation from your main screen
You can't unmute someone else's presentation
More options for Nikolay Yankov
Zoom in
Open in new window
Enter Full Screen
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Pin Aneliya Angelova to your main screen
Mute Aneliya Angelova's microphone
More options for Aneliya Angelova
Aneliya Angelova
Pin Steliyan Georgiev to your main screen
You can't unmute someone else
More options for Steliyan Georgiev
Steliyan Georgiev
Pin Stefka Stoyanova to your main screen
You can't unmute someone else
More options for Nikolay Nikolov
Stefka Stoyanova
User profile picture User profile picture 2 others
2 others
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
You’re continuously framed
Backgrounds and effects
More options for Lukas Kovalik
Lukas Kovalik
Others might see more of your background. Click to view your full video.
10:13
AM
Daily - Platform
Daily - Platform
Audio settings
Turn off microphone
Video settings
Turn off camera
Nikolay Yankov is presenting
Send a reaction
Turn on captions
Raise hand (ctrl + ⌘ + h)
More options
Leave call
Meeting details
Chat with everyone
Meeting tools...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Meet - Daily - Platform","depth":4,"bounds":{"left":0.27027926,"top":1.0,"width":0.016123671,"height":-0.051875472},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.27094415,"top":1.0,"width":0.004986702,"height":-0.051875472},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.27310506,"top":1.0,"width":0.010638298,"height":-0.086193085},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"Nikolay Yankov (Presenting, annotating)","depth":12,"bounds":{"left":0.30634972,"top":1.0,"width":0.08510638,"height":-0.072625756},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Nikolay Yankov (Presenting, annotating)","depth":13,"bounds":{"left":0.30634972,"top":1.0,"width":0.08510638,"height":-0.07342374},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"People","depth":15,"bounds":{"left":0.69481385,"top":1.0,"width":0.019614361,"height":-0.06424582},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"8","depth":22,"bounds":{"left":0.7081117,"top":1.0,"width":0.0023271276,"height":-0.072625756},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Take notes with Gemini","depth":14,"bounds":{"left":0.71708775,"top":1.0,"width":0.011968086,"height":-0.06424582},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Take notes with Gemini","depth":17,"bounds":{"left":0.7184175,"top":1.0,"width":0.043550532,"height":-0.072625756},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini","depth":22,"bounds":{"left":0.7330452,"top":1.0,"width":0.013464096,"height":-0.072625756},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Gemini","depth":21,"bounds":{"left":0.73204786,"top":1.0,"width":0.011303191,"height":-0.065043926},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Pop out this video More screens are more fun. Play this video while you do other things.","depth":15,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pop out this video","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"More screens are more fun. Play this video while you do other things.","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Unpin Nikolay Yankov's presentation from your main screen","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"You can't unmute someone else's presentation","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options for Nikolay Yankov","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Zoom in","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open in new window","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Enter Full Screen","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Pop out this video More screens are more fun. Play this video while you do other things.","depth":15,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pop out this video","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"More screens are more fun. Play this video while you do other things.","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Pin Aneliya Angelova to your main screen","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Mute Aneliya Angelova's microphone","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options for Aneliya Angelova","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Aneliya Angelova","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Pin Steliyan Georgiev to your main screen","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"You can't unmute someone else","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options for Steliyan Georgiev","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Steliyan Georgiev","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Pin Stefka Stoyanova to your main screen","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"You can't unmute someone else","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options for Nikolay Nikolov","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Stefka Stoyanova","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"User profile picture User profile picture 2 others","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"2 others","depth":13,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Pop out this video More screens are more fun. Play this video while you do other things.","depth":15,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pop out this video","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"More screens are more fun. Play this video while you do other things.","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"You’re continuously framed","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Backgrounds and effects","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options for Lukas Kovalik","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Lukas Kovalik","depth":17,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Others might see more of your background. Click to view your full video.","depth":14,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"10:13","depth":12,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AM","depth":12,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Daily - Platform","depth":12,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Daily - Platform","depth":15,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Audio settings","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Turn off microphone","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Video settings","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Turn off camera","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Nikolay Yankov is presenting","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Send a reaction","depth":12,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Turn on captions","depth":13,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Raise hand (ctrl + ⌘ + h)","depth":12,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"More options","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Leave call","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXButton","text":"Meeting details","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Chat with everyone","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Meeting tools","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false}]...
|
-8682764205109106750
|
-6570499528790967536
|
click
|
hybrid
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
Nikolay Yankov (Presenting, annotating)
Nikolay Yankov (Presenting, annotating)
People
8
Take notes with Gemini
Take notes with Gemini
Gemini
Gemini
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Unpin Nikolay Yankov's presentation from your main screen
You can't unmute someone else's presentation
More options for Nikolay Yankov
Zoom in
Open in new window
Enter Full Screen
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
Pin Aneliya Angelova to your main screen
Mute Aneliya Angelova's microphone
More options for Aneliya Angelova
Aneliya Angelova
Pin Steliyan Georgiev to your main screen
You can't unmute someone else
More options for Steliyan Georgiev
Steliyan Georgiev
Pin Stefka Stoyanova to your main screen
You can't unmute someone else
More options for Nikolay Nikolov
Stefka Stoyanova
User profile picture User profile picture 2 others
2 others
Pop out this video More screens are more fun. Play this video while you do other things.
Pop out this video
More screens are more fun. Play this video while you do other things.
You’re continuously framed
Backgrounds and effects
More options for Lukas Kovalik
Lukas Kovalik
Others might see more of your background. Click to view your full video.
10:13
AM
Daily - Platform
Daily - Platform
Audio settings
Turn off microphone
Video settings
Turn off camera
Nikolay Yankov is presenting
Send a reaction
Turn on captions
Raise hand (ctrl + ⌘ + h)
More options
Leave call
Meeting details
Chat with everyone
Meeting tools
WindowFV faVsco.js?9 JY-20725-handle-HS-search-rate-limitProiect vC) AutomatedReportGenerated.onp© PlaybackController.php xC) SubscrintionControfinal class PlaybackController extends FrontenoC) TeamA AutomationC) TeamA ConteytConФ TeamController.phpc) ToaminciahtcContr© TranscriptionContrc(C) TranclationControllc usercontroller.pnp© VocabularyControll>@ Auth-customerapl›J Internav D Kioskeams© ActivityController.pAutomatedReportsic) DashboardControllec) ImpersonationContC) MediaPinelinecont0 OraanizationsContrC PartnersController.C) ProfileController.ohC)SearchController.of>• Settings>• Telephonyv M Wehhook> M Huhsnot>D IntegrationAppSub:© ActivityProviderCor(C) ActivitvTranccrintio© BaseController.php© CalendarController,© ReportController.phSoftphoneWebhool.AhstractController.ong@ CommentContextinter© ConferencesOptInOutcc) controller.phpExportController.phpc)FrontendController.oh() FrontendControllerTra@ GeocodinaController.n(C) HealthCheckControllei@ LiveCoachController.p(C) Missina TeamControlle(C) MobileController.ohvC) NotificationController.t© NotificationProviderCoYe) PlavbackControlier ohg DlavlistController nhn(C) PucherController nhng SlackController nhnpudLlc tunction snow Activlcy$data = Fractal::createQ->item(SacuLVIEynStnancfonmen-scetConcumen/Sucer->serzal1zewichnew Jsonserallzer->cOArrayo"Sdata['playbackData']['masterPlaylist']if (! isset(Sdata['playbackData']['traoSdatal'playbackData']['tracks'] = 1* Sendina 'plaubackVisited' event to* response to the user e.a.nften thedeferdfnO => Sthis->nlanhatService->tr:user: Suserevent: 'nlavbackVisited'payload: ['activityId' => $activity->'activityUuid' => Sactivit)->alwaysO:Ireturn $this->renderCO'playbackData' =>['activity' => Sdatal'playbackDa'favorited' => Sdata['favorite'subscribed' => Sdatal'subscril'VleW' => Svleworivate function getPreloadedPlavistlActl* athrows AuthorizationExcentionGET Inlavlict[activitv) m2u8 Inlavlictlpublic function playlist(Activity SactivityNew Tabe Jy 20820 es reindex stream mode4 [JY-20725] [HubSpot] Optimise Cl• JY-20725 add HS rate limit handlilPipelines - jiminny/app• Pull requests • jiminny/appWY-207731 User Pilot not receivinJY-20773 fix user pilot tracking ofiNY-20776 Automated report - seTypeError: League|Flysystem\FilesA TypeError: League\Flysystem\FilesPlatform Sprint 3 02 - Platform Te• JY-20625 | JY-20742 | MCP POC kG LLM pricing overview with the mo:G olanhat - Gooale SearcholannalGoogle GemintI'm on page"<tabTitle>JY-20361: Addcall scores in Panorama bystelivan-</tao lites> with"‹selection>@@ -4,7 +4,7...show thinkingHere is a summary or the changes in theprovided code diff.Al Calll Scorind Additions• Pre-computed Scores: Added strictinstructions to treat the ai_call_scoreas ore-comouree oala. Ine system musInterorer ano summarize the nrovidodscores rather than inventing orrecalculating them• Context Breakdown: * Short Context:Now disolavs the Al Scorecard name andthe overall Al Score (the averaae of per-rule scores).• Full Context: Details theai call score object, breakingcown individual rules. Fach ruleincludes a title, prompt, a 1-5 score, abriet justitication, and up to threetimestamoed soeaker reterencesO Enter a prompt for GeminiPro vYour Jiminny chats aren't used to improve our models. Gemini oAlrcan make micnkec col daliale cnockt Yottr drivaeu wdomSummarize pageGoogleolannaluAl ModeShort videosReviewsSponsored resultPlanhat - Customer Success PlatformManage CS in one place. Gain visibility, reduce churn, automate key journeys. Visibility, predictionautomation—one CS platform to run it allAll-in-one CS PlatformUnify CS ops in one platform. Predict churn. spot expansion. and automate iournevs.Churn Prediction and PlaybooksScale vour CS team without the chaos—see more, know more, do more in one platform.Advance Al FeaturesRuilt in Al tor advance tacks Get more done taster with AlIMultiple IntearationsEasilv connect with vour tech stack Easilv connect with existina toolsAccurate ReportingEasily access data via dashboards Easily visualize customer dataHide sponsored result^Planhat.httns://ww.nlanhat.comPlanhatA customer platform where commercial teams work better. We build and deploy frontier technologies-intearatina data. Al and collaboration—that empower ..AhounA workspace that allows you to monitor and manage your ….CaoturacExolore Planhat's powerful features. Unify vour data …•$0 llsuppon Dally• In 4h 47 m100% 5• lue 1z May 10-13.41NORRMALMISee nhotosPlanhat50 ***** 1 review ©Software comnanv in Stockholm Swedenl• Wehsite* Directions* ReviewsW SaveL ShareainsrodilineallllonanAddress: Sveavägen 17, 111 57 Stockholm, SwedenHours: Onen 24 hoursSuggest an edit • Own this business?Add missing informationAdd nlaces nhone numberFrom P anhar"Planhat is helping hundreds of modern technology companies worldwidecenter their business around their customers in order to maximizecuctomer cuccoce and cuctomor lifotime value "# Send to your phoneProfilesYoullubeLinkedinX aiwitter)...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22559
|
973
|
25
|
2026-05-12T07:13:50.573045+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570030573_m2.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
You left the meeting
You left the meeting
Rejoin
Rejoin
Return to home screen
Return to home screen
How was the audio and video?
How was the audio and video?
Rate the meeting 1 star out of 5.
Rate the meeting 2 stars out of 5.
Rate the meeting 3 stars out of 5.
Rate the meeting 4 stars out of 5.
Rate the meeting 5 stars out of 5.
Very bad
Very good
Feedback
Feedback
58
Returning to home screen
Returning to home screen in 60 seconds.
Background is no longer replaced...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Meet - Daily - Platform","depth":4,"bounds":{"left":0.27027926,"top":1.0,"width":0.016123671,"height":-0.051875472},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.27094415,"top":1.0,"width":0.004986702,"height":-0.051875472},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.27310506,"top":1.0,"width":0.010638298,"height":-0.086193085},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You left the meeting","depth":10,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You left the meeting","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Rejoin","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXStaticText","text":"Rejoin","depth":13,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Return to home screen","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Return to home screen","depth":13,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How was the audio and video?","depth":10,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How was the audio and video?","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Rate the meeting 1 star out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 2 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 3 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 4 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 5 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Very bad","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Very good","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Feedback","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Feedback","depth":13,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"58","depth":12,"bounds":{"left":0.2997008,"top":1.0,"width":0.005319149,"height":-0.083798885},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Returning to home screen","depth":12,"bounds":{"left":0.31432846,"top":1.0,"width":0.0546875,"height":-0.083798885},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Returning to home screen in 60 seconds.","depth":8,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Background is no longer replaced","depth":8,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-5487692783768348400
|
705405106535182090
|
click
|
accessibility
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
You left the meeting
You left the meeting
Rejoin
Rejoin
Return to home screen
Return to home screen
How was the audio and video?
How was the audio and video?
Rate the meeting 1 star out of 5.
Rate the meeting 2 stars out of 5.
Rate the meeting 3 stars out of 5.
Rate the meeting 4 stars out of 5.
Rate the meeting 5 stars out of 5.
Very bad
Very good
Feedback
Feedback
58
Returning to home screen
Returning to home screen in 60 seconds.
Background is no longer replaced...
|
22557
|
NULL
|
NULL
|
NULL
|
|
22560
|
972
|
47
|
2026-05-12T07:13:51.616881+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570031616_m1.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
You left the meeting
You left the meeting
Rejoin
Rejoin
Return to home screen
Return to home screen
How was the audio and video?
How was the audio and video?
Rate the meeting 1 star out of 5.
Rate the meeting 2 stars out of 5.
Rate the meeting 3 stars out of 5.
Rate the meeting 4 stars out of 5....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Meet - Daily - Platform","depth":4,"bounds":{"left":0.0,"top":0.072222225,"width":0.033680554,"height":0.045555554},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.0013888889,"top":0.072222225,"width":0.010416667,"height":0.016666668},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.005902778,"top":0.12,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.7977778,"width":0.033680554,"height":0.043333333},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.8411111,"width":0.033680554,"height":0.038333334},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.0,"top":0.8794444,"width":0.033680554,"height":0.03888889},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.0,"top":0.91833335,"width":0.033680554,"height":0.038333334},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.95666665,"width":0.033680554,"height":0.043333333},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You left the meeting","depth":10,"bounds":{"left":0.4045139,"top":0.18333334,"width":0.22465278,"height":0.04888889},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You left the meeting","depth":11,"bounds":{"left":0.4045139,"top":0.18277778,"width":0.22465278,"height":0.050555557},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Rejoin","depth":11,"bounds":{"left":0.41493055,"top":0.27222222,"width":0.062152777,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXStaticText","text":"Rejoin","depth":13,"bounds":{"left":0.43229166,"top":0.28444445,"width":0.027430555,"height":0.020555556},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Return to home screen","depth":11,"bounds":{"left":0.4826389,"top":0.27222222,"width":0.13611111,"height":0.044444446},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Return to home screen","depth":13,"bounds":{"left":0.49930555,"top":0.28444445,"width":0.10277778,"height":0.020555556},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How was the audio and video?","depth":10,"bounds":{"left":0.41284722,"top":0.39222223,"width":0.20833333,"height":0.046666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How was the audio and video?","depth":11,"bounds":{"left":0.41284722,"top":0.39444444,"width":0.15729167,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Rate the meeting 1 star out of 5.","depth":10,"bounds":{"left":0.41631943,"top":0.43888888,"width":0.033333335,"height":0.053333335},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 2 stars out of 5.","depth":10,"bounds":{"left":0.45833334,"top":0.43888888,"width":0.033333335,"height":0.053333335},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 3 stars out of 5.","depth":10,"bounds":{"left":0.5003472,"top":0.43888888,"width":0.033333335,"height":0.053333335},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 4 stars out of 5.","depth":10,"bounds":{"left":0.54236114,"top":0.43888888,"width":0.033333335,"height":0.053333335},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-6932076777605361764
|
5285003286854938246
|
visual_change
|
hybrid
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
You left the meeting
You left the meeting
Rejoin
Rejoin
Return to home screen
Return to home screen
How was the audio and video?
How was the audio and video?
Rate the meeting 1 star out of 5.
Rate the meeting 2 stars out of 5.
Rate the meeting 3 stars out of 5.
Rate the meeting 4 stars out of 5.
Firefox File Edit View•••@<→ CHistoryBookmarksProfilesToolsWindowHelp• =@ meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com{ Support Daily - in 4h 47 m A *100% C4 8• Tue 12 May 10:13:51(58)Returning to home screenYou left the meetingRejoinReturn to home screenHow was the audio and video?Very badVery goodСopy51 27m 56s1,02 GBFeedback...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22561
|
973
|
26
|
2026-05-12T07:13:51.906057+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570031906_m2.jpg...
|
Firefox
|
Meet - Daily - Platform — Work
|
1
|
meet.google.com/mie-gawc-dsi?authuser=lukas.kovali meet.google.com/mie-gawc-dsi?authuser=lukas.kovalik%40jiminny.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
You left the meeting
You left the meeting
Rejoin
Rejoin
Return to home screen
Return to home screen
How was the audio and video?
How was the audio and video?
Rate the meeting 1 star out of 5.
Rate the meeting 2 stars out of 5.
Rate the meeting 3 stars out of 5.
Rate the meeting 4 stars out of 5.
Rate the meeting 5 stars out of 5.
Very bad
Very good
Feedback
Feedback
57
Returning to home screen
Returning to home screen in 60 seconds.
Background is no longer replaced...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"Meet - Daily - Platform","depth":4,"bounds":{"left":0.27027926,"top":1.0,"width":0.016123671,"height":-0.051875472},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.27094415,"top":1.0,"width":0.004986702,"height":-0.051875472},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.27310506,"top":1.0,"width":0.010638298,"height":-0.086193085},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open Google Gemini (⌃X)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You left the meeting","depth":10,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You left the meeting","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Rejoin","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":true,"is_selected":false},{"role":"AXStaticText","text":"Rejoin","depth":13,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Return to home screen","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Return to home screen","depth":13,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"How was the audio and video?","depth":10,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How was the audio and video?","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Rate the meeting 1 star out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 2 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 3 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 4 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Rate the meeting 5 stars out of 5.","depth":10,"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Very bad","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Very good","depth":11,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Feedback","depth":11,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Feedback","depth":13,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"57","depth":12,"bounds":{"left":0.2997008,"top":1.0,"width":0.005319149,"height":-0.083798885},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Returning to home screen","depth":12,"bounds":{"left":0.31432846,"top":1.0,"width":0.0546875,"height":-0.083798885},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Returning to home screen in 60 seconds.","depth":8,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Background is no longer replaced","depth":8,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-7873038905984180355
|
5316528175008886410
|
click
|
accessibility
|
NULL
|
Meet - Daily - Platform
Close tab
New Tab
Open Goo Meet - Daily - Platform
Close tab
New Tab
Open Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
Customize sidebar
You left the meeting
You left the meeting
Rejoin
Rejoin
Return to home screen
Return to home screen
How was the audio and video?
How was the audio and video?
Rate the meeting 1 star out of 5.
Rate the meeting 2 stars out of 5.
Rate the meeting 3 stars out of 5.
Rate the meeting 4 stars out of 5.
Rate the meeting 5 stars out of 5.
Very bad
Very good
Feedback
Feedback
57
Returning to home screen
Returning to home screen in 60 seconds.
Background is no longer replaced...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22592
|
975
|
6
|
2026-05-12T07:14:47.909300+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570087909_m2.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Skip to main content
Skip to main content
Accessibility help
Accessibility help
Accessibility feedback
Accessibility feedback
Go to Google Home
planhat
planhat
Clear
Search by voice
Search by image
Search
Google apps
Google Account: [EMAIL]
AI Mode
AI Mode
All
All
Images
Images
News
News
Videos
Videos
Forums
Forums
Short videos
Short videos
More filters
More
Tools
Tools
Add Reviews
Reviews
Search Results
Search Results...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.042220745,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5171588,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.5442937,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Skip to main content","depth":7,"bounds":{"left":0.46243352,"top":0.0981644,"width":0.03656915,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Skip to main content","depth":8,"bounds":{"left":0.46791887,"top":0.101356745,"width":0.025598405,"height":0.028731046},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Accessibility help","depth":7,"bounds":{"left":0.46243352,"top":0.0981644,"width":0.03656915,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Accessibility help","depth":8,"bounds":{"left":0.4680851,"top":0.101356745,"width":0.025265958,"height":0.028731046},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Accessibility feedback","depth":7,"bounds":{"left":0.46243352,"top":0.12051077,"width":0.03656915,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Accessibility feedback","depth":8,"bounds":{"left":0.4680851,"top":0.123703115,"width":0.025265958,"height":0.028731046},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Go to Google Home","depth":10,"bounds":{"left":0.48337767,"top":0.08060654,"width":0.030585106,"height":0.026336791},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXComboBox","text":"planhat","depth":9,"bounds":{"left":0.5355718,"top":0.07342378,"width":0.21875,"height":0.03990423},"on_screen":true,"value":"planhat","help_text":"","role_description":"combo box","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"planhat","depth":10,"bounds":{"left":0.5355718,"top":0.08619314,"width":0.017453458,"height":0.014365523},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Clear","depth":9,"bounds":{"left":0.7543218,"top":0.07342378,"width":0.015957447,"height":0.03990423},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search by voice","depth":9,"bounds":{"left":0.7719415,"top":0.083798885,"width":0.013297873,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search by image","depth":9,"bounds":{"left":0.78523934,"top":0.083798885,"width":0.013297873,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Search","depth":9,"bounds":{"left":0.79986703,"top":0.07342378,"width":0.01462766,"height":0.03990423},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Google apps","depth":9,"bounds":{"left":0.96392953,"top":0.07741421,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXMenuButton","text":"Google Account: lukas.kovalik@jiminny.com","depth":8,"bounds":{"left":0.97988695,"top":0.07741421,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"menu button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXLink","text":"AI Mode","depth":17,"bounds":{"left":0.53125,"top":0.12210695,"width":0.025265958,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"AI Mode","depth":19,"bounds":{"left":0.53523934,"top":0.13727055,"width":0.017287234,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"All","depth":17,"bounds":{"left":0.55651593,"top":0.12210695,"width":0.013131649,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"All","depth":20,"bounds":{"left":0.56050533,"top":0.13727055,"width":0.0051529254,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Images","depth":17,"bounds":{"left":0.5696476,"top":0.12210695,"width":0.023271276,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Images","depth":19,"bounds":{"left":0.57363695,"top":0.13727055,"width":0.015292553,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"News","depth":17,"bounds":{"left":0.5929189,"top":0.12210695,"width":0.019614361,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"News","depth":19,"bounds":{"left":0.5969083,"top":0.13727055,"width":0.011635638,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Videos","depth":17,"bounds":{"left":0.6125333,"top":0.12210695,"width":0.022107713,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Videos","depth":19,"bounds":{"left":0.6165226,"top":0.13727055,"width":0.01412899,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Forums","depth":17,"bounds":{"left":0.63464093,"top":0.12210695,"width":0.023769947,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Forums","depth":19,"bounds":{"left":0.63863033,"top":0.13727055,"width":0.015791224,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Short videos","depth":17,"bounds":{"left":0.6584109,"top":0.12210695,"width":0.033909574,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Short videos","depth":19,"bounds":{"left":0.66240025,"top":0.13727055,"width":0.025930852,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"More filters","depth":17,"bounds":{"left":0.69232047,"top":0.12210695,"width":0.024601065,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"More","depth":20,"bounds":{"left":0.69630986,"top":0.13727055,"width":0.010638298,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Tools","depth":16,"bounds":{"left":0.71692157,"top":0.12210695,"width":0.024767287,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"Tools","depth":18,"bounds":{"left":0.7209109,"top":0.13727055,"width":0.010804521,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Add Reviews","depth":14,"bounds":{"left":0.53523934,"top":0.16679968,"width":0.02825798,"height":0.028731046},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Reviews","depth":16,"bounds":{"left":0.5405585,"top":0.17478053,"width":0.01761968,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Search Results","depth":8,"bounds":{"left":0.4587766,"top":0.19952115,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Search Results","depth":9,"bounds":{"left":0.4587766,"top":0.19952115,"width":0.03158245,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-1973460085984354890
|
-2831911621274387566
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Skip to main content
Skip to main content
Accessibility help
Accessibility help
Accessibility feedback
Accessibility feedback
Go to Google Home
planhat
planhat
Clear
Search by voice
Search by image
Search
Google apps
Google Account: [EMAIL]
AI Mode
AI Mode
All
All
Images
Images
News
News
Videos
Videos
Forums
Forums
Short videos
Short videos
More filters
More
Tools
Tools
Add Reviews
Reviews
Search Results
Search Results...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22593
|
974
|
5
|
2026-05-12T07:14:49.709411+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570089709_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.09652778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.19861111,"height":0.10944445},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.09861111,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.20173611,"height":0.10944445},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"bounds":{"left":0.16527778,"top":0.19944444,"width":0.017708333,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"bounds":{"left":0.18715277,"top":0.19777778,"width":0.11215278,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"bounds":{"left":0.123958334,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"bounds":{"left":0.14618056,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"bounds":{"left":0.16840278,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"bounds":{"left":0.190625,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"bounds":{"left":0.21284722,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.15729167,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"bounds":{"left":0.18784723,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"bounds":{"left":0.22951388,"top":0.34777778,"width":0.13611111,"height":0.053333335},"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.109375,"top":0.34888887,"width":0.041666668,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"bounds":{"left":0.22951388,"top":0.35,"width":0.13229166,"height":0.049444444},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.34895834,"top":0.44555557,"width":0.027777778,"height":0.044444446},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.15868056,"top":0.4488889,"width":0.06458333,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.15381944,"top":0.50666666,"width":0.00069444446,"height":0.0011111111},"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.15381944,"top":0.5094444,"width":0.08576389,"height":0.026666667},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"bounds":{"left":0.13090278,"top":0.51111114,"width":0.2375,"height":0.19611111},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"bounds":{"left":0.13090278,"top":0.72555554,"width":0.24027778,"height":0.051666666},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"bounds":{"left":0.13090278,"top":0.8111111,"width":0.24027778,"height":0.053333335},"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"bounds":{"left":0.13090278,"top":0.81333333,"width":0.2204861,"height":0.049444444},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"bounds":{"left":0.15729167,"top":0.87666667,"width":0.11840278,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"bounds":{"left":0.15729167,"top":0.87666667,"width":0.20208333,"height":0.051666666},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.17986111,"top":0.9072222,"width":0.07569444,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"bounds":{"left":0.15729167,"top":0.90555555,"width":0.20104167,"height":0.09444445},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"bounds":{"left":0.2777778,"top":0.99222225,"width":0.03923611,"height":0.0077777505},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"bounds":{"left":0.15729167,"top":0.99222225,"width":0.18888889,"height":0.0077777505},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"bounds":{"left":0.15729167,"top":1.0,"width":0.12777779,"height":-0.062222242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"bounds":{"left":0.15729167,"top":1.0,"width":0.19131945,"height":-0.062222242},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"bounds":{"left":0.15729167,"top":1.0,"width":0.20104167,"height":-0.091111064},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-2483094747131329737
|
-2543683376231761262
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options...
|
22589
|
NULL
|
NULL
|
NULL
|
|
22594
|
974
|
6
|
2026-05-12T07:14:51.100965+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570091100_m1.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
8432176313194120725
|
-2579134922916660606
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22595
|
975
|
7
|
2026-05-12T07:14:51.086402+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570091086_m2.jpg...
|
Firefox
|
planhat - Google Search — Work
|
1
|
www.planhat.com
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"planhat - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"planhat - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.042220745,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5171588,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.5442937,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-7599196871440037926
|
-2831770885395858542
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
planhat - Google Search
planhat - Google Search
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said...
|
22592
|
NULL
|
NULL
|
NULL
|
|
22596
|
975
|
8
|
2026-05-12T07:14:53.975064+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570093975_m2.jpg...
|
Firefox
|
Planhat — Work
|
1
|
www.planhat.com
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
THE AGENTIC CUSTOMER PLATFORM
THE AGENTIC CUSTOMER PLATFORM
Deploy AI with confidence across the customer lifecycle
Deploy AI with confidence across the customer lifecycle
Planhat provides B2B enterprises with forward-deployed software and services that deliver lifelong revenue growth.
REQUEST A DEMO
REQUEST A DEMO
Solutions
Platform
Customers
Explore
Pricing
Pricing
LOG IN
LOG IN
REQUEST A DEMO
REQUEST A DEMO
TRUSTED BY LEADING COMPANIES WORLDWIDE DISCOVER OUR IMPACT
TRUSTED BY LEADING COMPANIES WORLDWIDE
DISCOVER OUR IMPACT
Drive real-world outcomes by automating commercial processes
Drive real-world outcomes by automating commercial processes
A comprehensive platform that empowers artificial and human intelligence to collaborate seamlessly across the entire customer lifecycle.
Assemble your GTM under one roof
Assemble your GTM under one roof
Arm your go-to-market agents, teams and tools with rich context that turns post-sale truth into pre-sales intelligence.
Deploy agents with confidence
Deploy agents with confidence
Delegate processes to tightly governed agents which execute with greater speed and precision than humanly possible.
Stay in-the-loop
Stay in-the-loop
Give the your team the capability to leverage, supervise and enhance agentic outputs at scale.
THE LEADING CUSTOMER PLATFORM The next generation has arrived Planhat is recognized by both industry analysts and customers as a global leader in deploying AI-led automation for go-to-market teams.
THE LEADING CUSTOMER PLATFORM
THE LEADING CUSTOMER PLATFORM
The next generation has arrived
Planhat is recognized by both industry analysts and customers as a global leader in deploying AI-led automation for go-to-market teams.
880+ reviews
G2 Momentum Grid and Leader in 9 Categories.
Magic Quadrant™ for Customer Success Management Platforms
MarketScape for Worldwide Customer Success Applications
PLATFORM
Deploy with confidence
Deploy with
confidence
Planhat is architected to equip agents and humans with everything they need to execute with clarity and control.
Context
TRAINING
Collaboration
EXECUTION
Agents
INPUT
CONTEXT
Planhat constructs a comprehensive living model of your customers and commercial operations spanning traditional CRM data, rich time series data, SOPs and external telemetry.
APPLICATIONS One platform, built to power your reality
APPLICATIONS
One platform, built to power your reality
One platform, built to power your reality
Deploy Planhat as a CSP, CRM, PSA or a powerful all-in one configured to put artificial and human intelligence to work across the full lifecycle.
CUSTOMER SUCCESS
SALES & GTM
PROFESSIONAL SERVICES
CUSTOM BUILD
Dynamically score health
Dynamically score health
Use analytics along with input from humans and AI to deploy adaptable scoring that gauges whether customers are truly thriving or failing.
Capture every touchpoint
Capture every touchpoint
Planhat automatically gathers transcripts, emails, tickets and every other interaction—allowing you to focus on execution, not admin.
Sync seamlessly with Sales
Sync seamlessly with Sales
Use Planhat together with all major CRMs to ensure you get the full picture and move in sync with your entire organization.
SERVICES
SERVICES
Don't buy hypotheticals. Buy outcomes.
Don't buy hypotheticals.
Buy outcomes.
Planhat dispatches forward-deployed teams that work shoulder-to-shoulder with you to deliver real-world outcomes.
34 % MORE CUSTOMERS PER CSM
34
%
MORE CUSTOMERS PER CSM
+ 300 HOURS SAVED ANNUALLY PER PERSON
+
300
HOURS SAVED ANNUALLY PER PERSON
900 % SEAT EXPANSION
900
%
SEAT EXPANSION
21 % LESS CHURN
21
%
LESS CHURN
Operational
efficiency
Revenue
impact
FORWARD-DEPLOYED TEAMS AI Deployment Program Real-world commercial outcomes delivered at unprecedented speeds.
FORWARD-DEPLOYED TEAMS
FORWARD-DEPLOYED TEAMS
AI Deployment Program
AI Deployment Program
Real-world commercial outcomes delivered at unprecedented speeds.
PROCESS AUTOMATION Value Framework A system of processes architected to deliver commercial outcomes.
PROCESS AUTOMATION
PROCESS AUTOMATION
Value Framework
Value Framework
A system of processes architected to deliver commercial outcomes.
PLANHAT FOR Service-Intensive Software Continuously unlock the value your customers need. Planhat is a system of action that empowers you to define and track customer objectives, identify value opportunities, run collaborative cross-functional workflows, and report on ROI.
PLANHAT FOR
Service-Intensive Software
Service-Intensive Software
Continuously unlock the value your customers need. Planhat is a system of action that empowers you to define and track customer objectives, identify value opportunities, run collaborative cross-functional workflows, and report on ROI....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5171588,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.5442937,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"THE AGENTIC CUSTOMER PLATFORM","depth":8,"bounds":{"left":0.5299202,"top":0.39465284,"width":0.090259306,"height":0.014365523},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"THE AGENTIC CUSTOMER PLATFORM","depth":9,"bounds":{"left":0.5299202,"top":0.39584997,"width":0.090259306,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Deploy AI with confidence across the customer lifecycle","depth":7,"bounds":{"left":0.5299202,"top":0.42817238,"width":0.24667554,"height":0.105347164},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Deploy AI with confidence across the customer lifecycle","depth":8,"bounds":{"left":0.5299202,"top":0.42577812,"width":0.24617687,"height":0.110135674},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat provides B2B enterprises with forward-deployed software and services that deliver lifelong revenue growth.","depth":8,"bounds":{"left":0.5299202,"top":0.55387074,"width":0.14261968,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"REQUEST A DEMO","depth":8,"bounds":{"left":0.5299202,"top":0.6268954,"width":0.06050532,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"REQUEST A DEMO","depth":10,"bounds":{"left":0.5359042,"top":0.6360734,"width":0.043882977,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Solutions","depth":12,"bounds":{"left":0.59275264,"top":0.07462091,"width":0.020279255,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Platform","depth":12,"bounds":{"left":0.64444816,"top":0.07462091,"width":0.018284574,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Customers","depth":12,"bounds":{"left":0.69414896,"top":0.07462091,"width":0.023936171,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Explore","depth":12,"bounds":{"left":0.74950135,"top":0.07462091,"width":0.016456118,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Pricing","depth":10,"bounds":{"left":0.7947141,"top":0.06823623,"width":0.020611702,"height":0.026336791},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pricing","depth":12,"bounds":{"left":0.79737365,"top":0.07462091,"width":0.015292553,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"LOG IN","depth":10,"bounds":{"left":0.8372673,"top":0.06624102,"width":0.02825798,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LOG IN","depth":12,"bounds":{"left":0.84325135,"top":0.075418994,"width":0.016954787,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"REQUEST A DEMO","depth":10,"bounds":{"left":0.86884975,"top":0.06624102,"width":0.06067154,"height":0.030327214},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"REQUEST A DEMO","depth":12,"bounds":{"left":0.87483376,"top":0.075418994,"width":0.044049203,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"TRUSTED BY LEADING COMPANIES WORLDWIDE DISCOVER OUR IMPACT","depth":8,"bounds":{"left":0.5299202,"top":1.0,"width":0.39893618,"height":-0.031923413},"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TRUSTED BY LEADING COMPANIES WORLDWIDE","depth":10,"bounds":{"left":0.5299202,"top":1.0,"width":0.1171875,"height":-0.033120513},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"DISCOVER OUR IMPACT","depth":10,"bounds":{"left":0.86602396,"top":1.0,"width":0.057513297,"height":-0.033120513},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Drive real-world outcomes by automating commercial processes","depth":8,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Drive real-world outcomes by automating commercial processes","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"A comprehensive platform that empowers artificial and human intelligence to collaborate seamlessly across the entire customer lifecycle.","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Assemble your GTM under one roof","depth":9,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Assemble your GTM under one roof","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Arm your go-to-market agents, teams and tools with rich context that turns post-sale truth into pre-sales intelligence.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Deploy agents with confidence","depth":9,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Deploy agents with confidence","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Delegate processes to tightly governed agents which execute with greater speed and precision than humanly possible.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Stay in-the-loop","depth":9,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Stay in-the-loop","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Give the your team the capability to leverage, supervise and enhance agentic outputs at scale.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"THE LEADING CUSTOMER PLATFORM The next generation has arrived Planhat is recognized by both industry analysts and customers as a global leader in deploying AI-led automation for go-to-market teams.","depth":8,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"THE LEADING CUSTOMER PLATFORM","depth":10,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"THE LEADING CUSTOMER PLATFORM","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The next generation has arrived","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat is recognized by both industry analysts and customers as a global leader in deploying AI-led automation for go-to-market teams.","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"880+ reviews","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"G2 Momentum Grid and Leader in 9 Categories.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Magic Quadrant™ for Customer Success Management Platforms","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MarketScape for Worldwide Customer Success Applications","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"PLATFORM","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Deploy with confidence","depth":8,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Deploy with","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"confidence","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat is architected to equip agents and humans with everything they need to execute with clarity and control.","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"TRAINING","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Collaboration","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"EXECUTION","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Agents","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"INPUT","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CONTEXT","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat constructs a comprehensive living model of your customers and commercial operations spanning traditional CRM data, rich time series data, SOPs and external telemetry.","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"APPLICATIONS One platform, built to power your reality","depth":8,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"APPLICATIONS","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"One platform, built to power your reality","depth":9,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"One platform, built to power your reality","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Deploy Planhat as a CSP, CRM, PSA or a powerful all-in one configured to put artificial and human intelligence to work across the full lifecycle.","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CUSTOMER SUCCESS","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SALES & GTM","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"PROFESSIONAL SERVICES","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CUSTOM BUILD","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Dynamically score health","depth":10,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dynamically score health","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Use analytics along with input from humans and AI to deploy adaptable scoring that gauges whether customers are truly thriving or failing.","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Capture every touchpoint","depth":11,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Capture every touchpoint","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat automatically gathers transcripts, emails, tickets and every other interaction—allowing you to focus on execution, not admin.","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Sync seamlessly with Sales","depth":11,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Sync seamlessly with Sales","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Use Planhat together with all major CRMs to ensure you get the full picture and move in sync with your entire organization.","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"SERVICES","depth":8,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SERVICES","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Don't buy hypotheticals. Buy outcomes.","depth":8,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Don't buy hypotheticals.","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Buy outcomes.","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Planhat dispatches forward-deployed teams that work shoulder-to-shoulder with you to deliver real-world outcomes.","depth":9,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"34 % MORE CUSTOMERS PER CSM","depth":8,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"34","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"%","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MORE CUSTOMERS PER CSM","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"+ 300 HOURS SAVED ANNUALLY PER PERSON","depth":8,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"+","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"300","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HOURS SAVED ANNUALLY PER PERSON","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"900 % SEAT EXPANSION","depth":8,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"900","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"%","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"SEAT EXPANSION","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"21 % LESS CHURN","depth":8,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"21","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"%","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"LESS CHURN","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Operational","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"efficiency","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Revenue","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"impact","depth":10,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"FORWARD-DEPLOYED TEAMS AI Deployment Program Real-world commercial outcomes delivered at unprecedented speeds.","depth":9,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"FORWARD-DEPLOYED TEAMS","depth":11,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"FORWARD-DEPLOYED TEAMS","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Deployment Program","depth":11,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Deployment Program","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Real-world commercial outcomes delivered at unprecedented speeds.","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"PROCESS AUTOMATION Value Framework A system of processes architected to deliver commercial outcomes.","depth":9,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"PROCESS AUTOMATION","depth":11,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"PROCESS AUTOMATION","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Value Framework","depth":11,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Value Framework","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"A system of processes architected to deliver commercial outcomes.","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"PLANHAT FOR Service-Intensive Software Continuously unlock the value your customers need. Planhat is a system of action that empowers you to define and track customer objectives, identify value opportunities, run collaborative cross-functional workflows, and report on ROI.","depth":9,"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"PLANHAT FOR","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Service-Intensive Software","depth":10,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Service-Intensive Software","depth":11,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Continuously unlock the value your customers need. Planhat is a system of action that empowers you to define and track customer objectives, identify value opportunities, run collaborative cross-functional workflows, and report on ROI.","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
1652166333154523446
|
-2831915058053269357
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
THE AGENTIC CUSTOMER PLATFORM
THE AGENTIC CUSTOMER PLATFORM
Deploy AI with confidence across the customer lifecycle
Deploy AI with confidence across the customer lifecycle
Planhat provides B2B enterprises with forward-deployed software and services that deliver lifelong revenue growth.
REQUEST A DEMO
REQUEST A DEMO
Solutions
Platform
Customers
Explore
Pricing
Pricing
LOG IN
LOG IN
REQUEST A DEMO
REQUEST A DEMO
TRUSTED BY LEADING COMPANIES WORLDWIDE DISCOVER OUR IMPACT
TRUSTED BY LEADING COMPANIES WORLDWIDE
DISCOVER OUR IMPACT
Drive real-world outcomes by automating commercial processes
Drive real-world outcomes by automating commercial processes
A comprehensive platform that empowers artificial and human intelligence to collaborate seamlessly across the entire customer lifecycle.
Assemble your GTM under one roof
Assemble your GTM under one roof
Arm your go-to-market agents, teams and tools with rich context that turns post-sale truth into pre-sales intelligence.
Deploy agents with confidence
Deploy agents with confidence
Delegate processes to tightly governed agents which execute with greater speed and precision than humanly possible.
Stay in-the-loop
Stay in-the-loop
Give the your team the capability to leverage, supervise and enhance agentic outputs at scale.
THE LEADING CUSTOMER PLATFORM The next generation has arrived Planhat is recognized by both industry analysts and customers as a global leader in deploying AI-led automation for go-to-market teams.
THE LEADING CUSTOMER PLATFORM
THE LEADING CUSTOMER PLATFORM
The next generation has arrived
Planhat is recognized by both industry analysts and customers as a global leader in deploying AI-led automation for go-to-market teams.
880+ reviews
G2 Momentum Grid and Leader in 9 Categories.
Magic Quadrant™ for Customer Success Management Platforms
MarketScape for Worldwide Customer Success Applications
PLATFORM
Deploy with confidence
Deploy with
confidence
Planhat is architected to equip agents and humans with everything they need to execute with clarity and control.
Context
TRAINING
Collaboration
EXECUTION
Agents
INPUT
CONTEXT
Planhat constructs a comprehensive living model of your customers and commercial operations spanning traditional CRM data, rich time series data, SOPs and external telemetry.
APPLICATIONS One platform, built to power your reality
APPLICATIONS
One platform, built to power your reality
One platform, built to power your reality
Deploy Planhat as a CSP, CRM, PSA or a powerful all-in one configured to put artificial and human intelligence to work across the full lifecycle.
CUSTOMER SUCCESS
SALES & GTM
PROFESSIONAL SERVICES
CUSTOM BUILD
Dynamically score health
Dynamically score health
Use analytics along with input from humans and AI to deploy adaptable scoring that gauges whether customers are truly thriving or failing.
Capture every touchpoint
Capture every touchpoint
Planhat automatically gathers transcripts, emails, tickets and every other interaction—allowing you to focus on execution, not admin.
Sync seamlessly with Sales
Sync seamlessly with Sales
Use Planhat together with all major CRMs to ensure you get the full picture and move in sync with your entire organization.
SERVICES
SERVICES
Don't buy hypotheticals. Buy outcomes.
Don't buy hypotheticals.
Buy outcomes.
Planhat dispatches forward-deployed teams that work shoulder-to-shoulder with you to deliver real-world outcomes.
34 % MORE CUSTOMERS PER CSM
34
%
MORE CUSTOMERS PER CSM
+ 300 HOURS SAVED ANNUALLY PER PERSON
+
300
HOURS SAVED ANNUALLY PER PERSON
900 % SEAT EXPANSION
900
%
SEAT EXPANSION
21 % LESS CHURN
21
%
LESS CHURN
Operational
efficiency
Revenue
impact
FORWARD-DEPLOYED TEAMS AI Deployment Program Real-world commercial outcomes delivered at unprecedented speeds.
FORWARD-DEPLOYED TEAMS
FORWARD-DEPLOYED TEAMS
AI Deployment Program
AI Deployment Program
Real-world commercial outcomes delivered at unprecedented speeds.
PROCESS AUTOMATION Value Framework A system of processes architected to deliver commercial outcomes.
PROCESS AUTOMATION
PROCESS AUTOMATION
Value Framework
Value Framework
A system of processes architected to deliver commercial outcomes.
PLANHAT FOR Service-Intensive Software Continuously unlock the value your customers need. Planhat is a system of action that empowers you to define and track customer objectives, identify value opportunities, run collaborative cross-functional workflows, and report on ROI.
PLANHAT FOR
Service-Intensive Software
Service-Intensive Software
Continuously unlock the value your customers need. Planhat is a system of action that empowers you to define and track customer objectives, identify value opportunities, run collaborative cross-functional workflows, and report on ROI....
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22597
|
974
|
7
|
2026-05-12T07:14:55.722410+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570095722_m1.jpg...
|
Firefox
|
Planhat — Work
|
1
|
www.planhat.com
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes"....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.09652778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.19861111,"height":0.10944445},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
7198205658042726380
|
-2831911622884213870
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes"....
|
22594
|
NULL
|
NULL
|
NULL
|
|
22637
|
974
|
26
|
2026-05-12T07:16:24.008094+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570184008_m1.jpg...
|
Firefox
|
Planhat — Work
|
1
|
www.planhat.com
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
9052336404017279817
|
-4245703323335019890
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira...
|
22634
|
NULL
|
NULL
|
NULL
|
|
22638
|
974
|
27
|
2026-05-12T07:16:27.041738+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570187041_m1.jpg...
|
Firefox
|
Planhat — Work
|
1
|
www.planhat.com
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
-2517039603172596306
|
-3380775110537792864
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22639
|
975
|
30
|
2026-05-12T07:16:26.790228+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570186790_m2.jpg...
|
Firefox
|
Planhat — Work
|
1
|
www.planhat.com
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.5171588,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.5442937,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
1465360905504670865
|
-2543683719829144942
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22640
|
975
|
31
|
2026-05-12T07:16:28.634517+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570188634_m2.jpg...
|
Firefox
|
Planhat — Work
|
1
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
9052336404017279817
|
-4245703323335019890
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira...
|
22639
|
NULL
|
NULL
|
NULL
|
|
22641
|
974
|
28
|
2026-05-12T07:16:28.609952+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570188609_m1.jpg...
|
Firefox
|
Planhat — Work
|
1
|
www.planhat.com
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
4811382151791849269
|
4457142791016475577
|
click
|
hybrid
|
NULL
|
New Tab
QuickTime PlayerFileEditViewWindowHelpPetk New Tab
QuickTime PlayerFileEditViewWindowHelpPetko KashinskiScreen shareChromeFileEditViewHistoryBookmarksProfilesTabWindowHelp(NGreetiScorecandrejf wilsonCall Arws.planhat.com/jiminny/apps?id=66ceb97643c2530bb32c8bb6АIKB• ChatPlayground Al...Jiminny - Calenda...M GMailMy Calendly - Eve...= PH New UI Login+ BackSearch Jiminny83 App Center& UP > PH UXE Al apps& Created by meG Recentiy most activeIntegrations8 AutomationsP Private apps0 EditorPa Runs8 Data(aholSupport Daily - in 4h 44 mJiminnM InboxGet Starting with J...=Nate R( Apps= AрChloe Onboarding.BuildirQUserp+ CX Journey SMB....100% <78 • Tue 12 May 10:16:28+ |%8• Mon 11 May 12:17New tr+f Work,0 +. Petko•• XIEnabled4 Webhook incomingA WebhookCompletedWebhook&Event details8bcea7d0-160b- @TriggeranythingOmsTriggered by& Support User (Removed)6 Huddle with Lukas KovalikEa Event log95 Al Notes: OffailQ 100% ~app_token*:"NX-9946e170",*data*: (7&+Leave...
|
22638
|
NULL
|
NULL
|
NULL
|
|
22642
|
975
|
32
|
2026-05-12T07:16:30.226903+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570190226_m2.jpg...
|
Firefox
|
Planhat — Work
|
1
|
ws.planhat.com
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false}]...
|
-2868272289771096989
|
-2443956565351368050
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22643
|
975
|
33
|
2026-05-12T07:16:33.030084+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570193030_m2.jpg...
|
Firefox
|
Planhat — Work
|
1
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
2575318264314049800
|
-2434949366096627058
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini...
|
22642
|
NULL
|
NULL
|
NULL
|
|
22644
|
975
|
34
|
2026-05-12T07:16:36.020780+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570196020_m2.jpg...
|
Firefox
|
Login — Work
|
1
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Welcome back","depth":9,"bounds":{"left":0.48603722,"top":0.30526736,"width":0.14494681,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Welcome back","depth":10,"bounds":{"left":0.48603722,"top":0.30925778,"width":0.070644945,"height":0.030726258},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with Google","depth":9,"bounds":{"left":0.48603722,"top":0.3754988,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with Google","depth":11,"bounds":{"left":0.5398936,"top":0.38028732,"width":0.04454787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with SAML SSO","depth":9,"bounds":{"left":0.48603722,"top":0.40422985,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with SAML SSO","depth":11,"bounds":{"left":0.5362367,"top":0.40901837,"width":0.051861703,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":10,"bounds":{"left":0.55634975,"top":0.44014364,"width":0.0043218085,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Email","depth":10,"bounds":{"left":0.48603722,"top":0.46727854,"width":0.011136968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"you@example.com","depth":10,"bounds":{"left":0.49019283,"top":0.48643255,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"you@example.com","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Password","depth":10,"bounds":{"left":0.48603722,"top":0.5215483,"width":0.020279255,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.5940825,"top":0.5215483,"width":0.036901597,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"••••••","depth":10,"bounds":{"left":0.49019283,"top":0.54070234,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"Password","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Log in","depth":9,"bounds":{"left":0.48603722,"top":0.57821226,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Log in","depth":11,"bounds":{"left":0.5521942,"top":0.5830008,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-493634218375676853
|
-2831913821102687342
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22645
|
974
|
29
|
2026-05-12T07:16:40.906544+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570200906_m1.jpg...
|
Firefox
|
Login — Work
|
1
|
ws.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-4925565648599463165
|
-2831911623152649326
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22646
|
974
|
30
|
2026-05-12T07:16:43.376911+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570203376_m1.jpg...
|
Firefox
|
Login — Work
|
1
|
ws.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.09652778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.15729167,"top":0.0,"width":0.19861111,"height":0.10944445},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.09861111,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.15729167,"top":0.11111111,"width":0.20173611,"height":0.10944445},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"bounds":{"left":0.16527778,"top":0.19944444,"width":0.017708333,"height":0.020555556},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"bounds":{"left":0.18715277,"top":0.19777778,"width":0.11215278,"height":0.022777777},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"bounds":{"left":0.123958334,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"bounds":{"left":0.14618056,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"bounds":{"left":0.16840278,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"bounds":{"left":0.190625,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"bounds":{"left":0.21284722,"top":0.25,"width":0.022222223,"height":0.035555556},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.15729167,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"bounds":{"left":0.18784723,"top":0.33444443,"width":0.027777778,"height":0.044444446},"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
8733677560677435636
|
-2831911622884213870
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit...
|
22645
|
NULL
|
NULL
|
NULL
|
|
22647
|
975
|
35
|
2026-05-12T07:16:43.349644+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570203349_m2.jpg...
|
Firefox
|
Login — Work
|
1
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google Loading
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Welcome back","depth":9,"bounds":{"left":0.48603722,"top":0.30526736,"width":0.14494681,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Welcome back","depth":10,"bounds":{"left":0.48603722,"top":0.30925778,"width":0.070644945,"height":0.030726258},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with Google Loading","depth":9,"bounds":{"left":0.48603722,"top":0.3754988,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with Google","depth":11,"bounds":{"left":0.5398936,"top":0.38028732,"width":0.04454787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with SAML SSO","depth":9,"bounds":{"left":0.48603722,"top":0.40422985,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with SAML SSO","depth":11,"bounds":{"left":0.5362367,"top":0.40901837,"width":0.051861703,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":10,"bounds":{"left":0.55634975,"top":0.44014364,"width":0.0043218085,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Email","depth":10,"bounds":{"left":0.48603722,"top":0.46727854,"width":0.011136968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"you@example.com","depth":10,"bounds":{"left":0.49019283,"top":0.48643255,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"you@example.com","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Password","depth":10,"bounds":{"left":0.48603722,"top":0.5215483,"width":0.020279255,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.5940825,"top":0.5215483,"width":0.036901597,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"••••••","depth":10,"bounds":{"left":0.49019283,"top":0.54070234,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"Password","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Log in","depth":9,"bounds":{"left":0.48603722,"top":0.57821226,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Log in","depth":11,"bounds":{"left":0.5521942,"top":0.5830008,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
6791598485010019321
|
-2831913821102687342
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google Loading
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
22644
|
NULL
|
NULL
|
NULL
|
|
22648
|
975
|
36
|
2026-05-12T07:16:45.086821+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570205086_m2.jpg...
|
Firefox
|
Sign in - Google Accounts — Work
|
1
|
accounts.google.com/v3/signin/accountchooser?clien accounts.google.com/v3/signin/accountchooser?client_id=1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com&redirect_uri=https%3A%2F%2Fapi.planhat.com%2Foauth2%2Fgoogle&response_type=code&scope=email&state=backlink%3D%2F%26ws%3Dtrue%26nonce%3D98e5dc1d37a377b73ed53f4504cc0cc0&dsh=S-1608494723%3A1778570203972862&o2v=2&service=lso&flowName=GeneralOAuthFlow&opparams=%253F&continue=https%3A%2F%2Faccounts.google.com%2Fsignin%2Foauth%2Fconsent%3Fauthuser%3Dunknown%26part%3DAJi8hAMhmYbVUs8yZK4zHz4UXpqvL3xQ9fvdfnlb5j7FPEpY2VxXwhDsnTorT-HT90Uq10gkYTnW_W_VBLZbi-cHAF6mVP6ROLrmd8JIWvHqfO6R5R-ec8qzfpYKmEhVeMqTakdQlVCokOwUHJYC90GdmGeGN5aDrfkpRUAmBVW4uyTGHcZo58zLBpGw-AMuwZS1Zoj07Kqqw3SHF3d3gfPMJAShUQjaX3j7kyKd_wMWc9saMsvGv-gCka7anrItpDc_Rvy-0stPRsJwkPRzkkj-c8LlbdhRiWlsFAym2tgaWPNRlwQQJibrJgku9KNh9soO3lJQ7SfY-d2tJ-ucRpUYIRT4kELx1hsOPdY0iOAh28FsnKWzH4yLVuB5s81qUa2Dzyl53DYXM7q0quaQDT-uCvkQylhfKTzIZ2nLbF29abYaOrMCvPwqMeQ74haMpGm0WNtBFxKptfKyiZBiDQeBNYRiJGSIpg%26flowName%3DGeneralOAuthFlow%26as%3DS-1608494723%253A1778570203972862%26client_id%3D1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com%26requestPath%3D%252Fsignin%252Foauth%252Fconsent%23&app_domain=https%3A%2F%2Fapi.planhat.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Sign in with Google
Choose an account
Choose an account
to continue to
Planhat
[EMAIL]
[EMAIL]
Integration Account [EMAIL] Signed out
Integration Account
[EMAIL]
Signed out
Use another account
Use another account
Before using this app, you can review Planhat’s
Privacy Policy
Privacy Policy
and
Terms of Service
Terms of Service
.
English (United States)
English (United States)
Help...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Sign in - Google Accounts","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Sign in - Google Accounts","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.045212764,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Sign in with Google","depth":11,"bounds":{"left":0.5724734,"top":0.35434955,"width":0.042054523,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Choose an account","depth":10,"bounds":{"left":0.56848407,"top":0.4640862,"width":0.15292554,"height":0.0415004},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Choose an account","depth":11,"bounds":{"left":0.56848407,"top":0.46249002,"width":0.12832446,"height":0.044692736},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to continue to","depth":11,"bounds":{"left":0.56848407,"top":0.5199521,"width":0.03474069,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Planhat","depth":11,"bounds":{"left":0.60322475,"top":0.5199521,"width":0.019614361,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXLink","text":"lukas.kovalik@jiminny.com","depth":14,"bounds":{"left":0.73204786,"top":0.42577812,"width":0.16356383,"height":0.0415004},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"lukas.kovalik@jiminny.com","depth":16,"bounds":{"left":0.7506649,"top":0.43934557,"width":0.0546875,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Integration Account integration-account@jiminny.com Signed out","depth":14,"bounds":{"left":0.73204786,"top":0.46807662,"width":0.16356383,"height":0.054269753},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Integration Account","depth":16,"bounds":{"left":0.7506649,"top":0.4792498,"width":0.04936835,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"integration-account@jiminny.com","depth":16,"bounds":{"left":0.7506649,"top":0.49760574,"width":0.070644945,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Signed out","depth":16,"bounds":{"left":0.8703458,"top":0.47805268,"width":0.019946808,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Use another account","depth":14,"bounds":{"left":0.73204786,"top":0.5231444,"width":0.16356383,"height":0.0415004},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Use another account","depth":16,"bounds":{"left":0.7506649,"top":0.5359138,"width":0.051861703,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Before using this app, you can review Planhat’s","depth":10,"bounds":{"left":0.73736703,"top":0.5989625,"width":0.099734046,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Privacy Policy","depth":10,"bounds":{"left":0.83710104,"top":0.5989625,"width":0.03158245,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Privacy Policy","depth":11,"bounds":{"left":0.83710104,"top":0.5989625,"width":0.03158245,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"and","depth":10,"bounds":{"left":0.8686835,"top":0.5989625,"width":0.009142287,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Terms of Service","depth":10,"bounds":{"left":0.73736703,"top":0.6149242,"width":0.03756649,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Terms of Service","depth":11,"bounds":{"left":0.73736703,"top":0.6149242,"width":0.03756649,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":10,"bounds":{"left":0.7749335,"top":0.6149242,"width":0.0013297872,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXComboBox","text":"English (United States)","depth":10,"bounds":{"left":0.55651593,"top":0.67398244,"width":0.061502658,"height":0.025538707},"on_screen":true,"value":"English (United States)","help_text":"","role_description":"combo box","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXStaticText","text":"","depth":12,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"English (United States)","depth":12,"bounds":{"left":0.5618351,"top":0.68076617,"width":0.04155585,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Help","depth":11,"bounds":{"left":0.8387633,"top":0.6675978,"width":0.016456118,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
8855053595188441098
|
-2831841528213161322
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Sign in with Google
Choose an account
Choose an account
to continue to
Planhat
[EMAIL]
[EMAIL]
Integration Account [EMAIL] Signed out
Integration Account
[EMAIL]
Signed out
Use another account
Use another account
Before using this app, you can review Planhat’s
Privacy Policy
Privacy Policy
and
Terms of Service
Terms of Service
.
English (United States)
English (United States)
Help...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22649
|
974
|
31
|
2026-05-12T07:16:46.046363+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570206046_m1.jpg...
|
Firefox
|
Sign in - Google Accounts — Work
|
1
|
accounts.google.com/v3/signin/accountchooser?clien accounts.google.com/v3/signin/accountchooser?client_id=1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com&redirect_uri=https%3A%2F%2Fapi.planhat.com%2Foauth2%2Fgoogle&response_type=code&scope=email&state=backlink%3D%2F%26ws%3Dtrue%26nonce%3D98e5dc1d37a377b73ed53f4504cc0cc0&dsh=S-1608494723%3A1778570203972862&o2v=2&service=lso&flowName=GeneralOAuthFlow&opparams=%253F&continue=https%3A%2F%2Faccounts.google.com%2Fsignin%2Foauth%2Fconsent%3Fauthuser%3Dunknown%26part%3DAJi8hAMhmYbVUs8yZK4zHz4UXpqvL3xQ9fvdfnlb5j7FPEpY2VxXwhDsnTorT-HT90Uq10gkYTnW_W_VBLZbi-cHAF6mVP6ROLrmd8JIWvHqfO6R5R-ec8qzfpYKmEhVeMqTakdQlVCokOwUHJYC90GdmGeGN5aDrfkpRUAmBVW4uyTGHcZo58zLBpGw-AMuwZS1Zoj07Kqqw3SHF3d3gfPMJAShUQjaX3j7kyKd_wMWc9saMsvGv-gCka7anrItpDc_Rvy-0stPRsJwkPRzkkj-c8LlbdhRiWlsFAym2tgaWPNRlwQQJibrJgku9KNh9soO3lJQ7SfY-d2tJ-ucRpUYIRT4kELx1hsOPdY0iOAh28FsnKWzH4yLVuB5s81qUa2Dzyl53DYXM7q0quaQDT-uCvkQylhfKTzIZ2nLbF29abYaOrMCvPwqMeQ74haMpGm0WNtBFxKptfKyiZBiDQeBNYRiJGSIpg%26flowName%3DGeneralOAuthFlow%26as%3DS-1608494723%253A1778570203972862%26client_id%3D1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com%26requestPath%3D%252Fsignin%252Foauth%252Fconsent%23&app_domain=https%3A%2F%2Fapi.planhat.com...
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores)....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Sign in - Google Accounts","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Sign in - Google Accounts","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-5097155351300601237
|
-2831770884053681262
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores)....
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22650
|
975
|
37
|
2026-05-12T07:16:46.029359+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570206029_m2.jpg...
|
Firefox
|
Sign in - Google Accounts — Work
|
1
|
accounts.google.com/v3/signin/accountchooser?clien accounts.google.com/v3/signin/accountchooser?client_id=1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com&redirect_uri=https%3A%2F%2Fapi.planhat.com%2Foauth2%2Fgoogle&response_type=code&scope=email&state=backlink%3D%2F%26ws%3Dtrue%26nonce%3D98e5dc1d37a377b73ed53f4504cc0cc0&dsh=S-1608494723%3A1778570203972862&o2v=2&service=lso&flowName=GeneralOAuthFlow&opparams=%253F&continue=https%3A%2F%2Faccounts.google.com%2Fsignin%2Foauth%2Fconsent%3Fauthuser%3Dunknown%26part%3DAJi8hAMhmYbVUs8yZK4zHz4UXpqvL3xQ9fvdfnlb5j7FPEpY2VxXwhDsnTorT-HT90Uq10gkYTnW_W_VBLZbi-cHAF6mVP6ROLrmd8JIWvHqfO6R5R-ec8qzfpYKmEhVeMqTakdQlVCokOwUHJYC90GdmGeGN5aDrfkpRUAmBVW4uyTGHcZo58zLBpGw-AMuwZS1Zoj07Kqqw3SHF3d3gfPMJAShUQjaX3j7kyKd_wMWc9saMsvGv-gCka7anrItpDc_Rvy-0stPRsJwkPRzkkj-c8LlbdhRiWlsFAym2tgaWPNRlwQQJibrJgku9KNh9soO3lJQ7SfY-d2tJ-ucRpUYIRT4kELx1hsOPdY0iOAh28FsnKWzH4yLVuB5s81qUa2Dzyl53DYXM7q0quaQDT-uCvkQylhfKTzIZ2nLbF29abYaOrMCvPwqMeQ74haMpGm0WNtBFxKptfKyiZBiDQeBNYRiJGSIpg%26flowName%3DGeneralOAuthFlow%26as%3DS-1608494723%253A1778570203972862%26client_id%3D1022862339368-ki4c97g1tmioldgdg8lag1m2niqvlcho.apps.googleusercontent.com%26requestPath%3D%252Fsignin%252Foauth%252Fconsent%23&app_domain=https%3A%2F%2Fapi.planhat.com...
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Sign in - Google Accounts","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Sign in - Google Accounts","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.045212764,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
6514073522262166208
|
-2831911622079432046
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Sign in - Google Accounts
Sign in - Google Accounts
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
22648
|
NULL
|
NULL
|
NULL
|
|
22651
|
975
|
38
|
2026-05-12T07:16:48.128581+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570208128_m2.jpg...
|
Firefox
|
Planhat — Work
|
1
|
ws.planhat.com/login?ga=1
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read apis.google.com...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read apis.google.com","depth":5,"bounds":{"left":0.46010637,"top":0.9876297,"width":0.038397606,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-5280693668421536418
|
-2831913822176429418
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read apis.google.com...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22652
|
975
|
39
|
2026-05-12T07:16:51.138262+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570211138_m2.jpg...
|
Firefox
|
Planhat — Work
|
1
|
ws.planhat.com/jiminny
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
5710214587205052025
|
-2831913822176429418
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Planhat
Planhat
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page...
|
22651
|
NULL
|
NULL
|
NULL
|
|
22653
|
975
|
40
|
2026-05-12T07:16:54.162847+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570214162_m2.jpg...
|
Firefox
|
Home — Work
|
1
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Home
Home
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Home","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Home","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.010139627,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"L","depth":11,"bounds":{"left":0.9730718,"top":0.06304868,"width":0.002493351,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Lukas","depth":10,"bounds":{"left":0.9808843,"top":0.06304868,"width":0.012300532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.09377494,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.09936153,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Data Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.11931365,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Data Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.12490024,"width":0.027925532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":12,"bounds":{"left":0.46426198,"top":0.14485236,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Calendar","depth":14,"bounds":{"left":0.47623006,"top":0.15043895,"width":0.01861702,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Notifications","depth":12,"bounds":{"left":0.46426198,"top":0.17039107,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Notifications","depth":14,"bounds":{"left":0.47623006,"top":0.17597765,"width":0.026263298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"More","depth":12,"bounds":{"left":0.46426198,"top":0.19592977,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"More","depth":14,"bounds":{"left":0.47623006,"top":0.20151636,"width":0.010638298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sections","depth":13,"bounds":{"left":0.46426198,"top":0.23264167,"width":0.07114362,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":15,"bounds":{"left":0.46559176,"top":0.23782921,"width":0.01662234,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CS Day-to-day","depth":16,"bounds":{"left":0.47822472,"top":0.26456505,"width":0.030751329,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚀 Getting started Guide","depth":18,"bounds":{"left":0.47822472,"top":0.29010376,"width":0.049700797,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🪬 Just CS Data","depth":18,"bounds":{"left":0.47822472,"top":0.3140463,"width":0.032081116,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"👉 Daily Operations","depth":18,"bounds":{"left":0.47822472,"top":0.33798882,"width":0.03956117,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🗓️ Weekly prep","depth":18,"bounds":{"left":0.47822472,"top":0.36193135,"width":0.03125,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🤑 Renewals and Upsell","depth":18,"bounds":{"left":0.47822472,"top":0.3858739,"width":0.048204787,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-2672622827034840214
|
-2831911622079432046
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Home
Home
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22654
|
975
|
41
|
2026-05-12T07:16:57.209914+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570217209_m2.jpg...
|
Firefox
|
Content Explorer — Work
|
1
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab
docs
.
...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.02925532,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"L","depth":11,"bounds":{"left":0.9730718,"top":0.06304868,"width":0.002493351,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Lukas","depth":10,"bounds":{"left":0.9808843,"top":0.06304868,"width":0.012300532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.09377494,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.09936153,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Data Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.11931365,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Data Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.12490024,"width":0.027925532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":12,"bounds":{"left":0.46426198,"top":0.14485236,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Calendar","depth":14,"bounds":{"left":0.47623006,"top":0.15043895,"width":0.01861702,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Notifications","depth":12,"bounds":{"left":0.46426198,"top":0.17039107,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Notifications","depth":14,"bounds":{"left":0.47623006,"top":0.17597765,"width":0.026263298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"More","depth":12,"bounds":{"left":0.46426198,"top":0.19592977,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"More","depth":14,"bounds":{"left":0.47623006,"top":0.20151636,"width":0.010638298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sections","depth":13,"bounds":{"left":0.46426198,"top":0.23264167,"width":0.07114362,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":15,"bounds":{"left":0.46559176,"top":0.23782921,"width":0.01662234,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CS Day-to-day","depth":16,"bounds":{"left":0.47822472,"top":0.26456505,"width":0.030751329,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚀 Getting started Guide","depth":18,"bounds":{"left":0.47822472,"top":0.29010376,"width":0.049700797,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🪬 Just CS Data","depth":18,"bounds":{"left":0.47822472,"top":0.3140463,"width":0.032081116,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"👉 Daily Operations","depth":18,"bounds":{"left":0.47822472,"top":0.33798882,"width":0.03956117,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🗓️ Weekly prep","depth":18,"bounds":{"left":0.47822472,"top":0.36193135,"width":0.03125,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🤑 Renewals and Upsell","depth":18,"bounds":{"left":0.47822472,"top":0.3858739,"width":0.048204787,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚨 Risk and Churn Analytics","depth":18,"bounds":{"left":0.47822472,"top":0.40981644,"width":0.05718085,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Implementation","depth":16,"bounds":{"left":0.47822472,"top":0.44972068,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Impl Projects","depth":18,"bounds":{"left":0.47822472,"top":0.47525936,"width":0.026761968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Trial Opps (Under Review)","depth":18,"bounds":{"left":0.47822472,"top":0.49920192,"width":0.054022606,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Stoyan's clients","depth":18,"bounds":{"left":0.47822472,"top":0.5231444,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":15,"bounds":{"left":0.54355055,"top":0.09537111,"width":0.046043884,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":17,"bounds":{"left":0.55285907,"top":0.0981644,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon For you","depth":16,"bounds":{"left":0.5944149,"top":0.08739027,"width":0.028091755,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"For you","depth":18,"bounds":{"left":0.6043883,"top":0.0981644,"width":0.015458777,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Sections","depth":16,"bounds":{"left":0.62317157,"top":0.08739027,"width":0.030585106,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":18,"bounds":{"left":0.633145,"top":0.0981644,"width":0.017952127,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Pages","depth":16,"bounds":{"left":0.65442157,"top":0.08739027,"width":0.025265958,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pages","depth":18,"bounds":{"left":0.664395,"top":0.0981644,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Archived","depth":16,"bounds":{"left":0.6803524,"top":0.08739027,"width":0.031083776,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Archived","depth":18,"bounds":{"left":0.6903258,"top":0.0981644,"width":0.018450798,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":14,"bounds":{"left":0.84890294,"top":0.09377494,"width":0.036402926,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":16,"bounds":{"left":0.85837764,"top":0.0981644,"width":0.018118352,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"Search...","depth":16,"bounds":{"left":0.55502,"top":0.12809257,"width":0.05086436,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"No content to display","depth":19,"bounds":{"left":0.6924867,"top":0.56264967,"width":0.043882977,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace","depth":19,"bounds":{"left":0.6549202,"top":0.5802075,"width":0.119015954,"height":0.027533919},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":18,"bounds":{"left":0.6961436,"top":0.6209098,"width":0.03656915,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":20,"bounds":{"left":0.7056183,"top":0.6256983,"width":0.018284574,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Content Explorer","depth":18,"bounds":{"left":0.56981385,"top":0.91660017,"width":0.050033245,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":20,"bounds":{"left":0.58111703,"top":0.9261772,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.62117684,"top":0.91660017,"width":0.118351065,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.63248,"top":0.9261772,"width":0.10305851,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.7408577,"top":0.91660017,"width":0.11818484,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.7521609,"top":0.9261772,"width":0.10289229,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Explore Keynote","depth":15,"bounds":{"left":0.9243683,"top":0.3292099,"width":0.0390625,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Last edited","depth":16,"bounds":{"left":0.89744014,"top":0.36871508,"width":0.021110373,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Active Renewal Opps","depth":18,"bounds":{"left":0.9054189,"top":0.3934557,"width":0.04637633,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2 minutes ago","depth":18,"bounds":{"left":0.9659242,"top":0.39385474,"width":0.02443484,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Existing Clients","depth":18,"bounds":{"left":0.9054189,"top":0.41660017,"width":0.03507314,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4169992,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"My Renewals next 2Qs","depth":18,"bounds":{"left":0.9054189,"top":0.43974462,"width":0.04338431,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.44014364,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OD Tasks","depth":18,"bounds":{"left":0.9054189,"top":0.46288908,"width":0.018118352,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4632881,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"First 30 days to adoption","depth":18,"bounds":{"left":0.9054189,"top":0.48603353,"width":0.04737367,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20 minutes ago","depth":18,"bounds":{"left":0.96359706,"top":0.48643255,"width":0.026761968,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Explore community templates","depth":16,"bounds":{"left":0.89744014,"top":0.5203512,"width":0.05651596,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Page templates","depth":18,"bounds":{"left":0.90807843,"top":0.5510774,"width":0.029421542,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Section templates","depth":18,"bounds":{"left":0.90807843,"top":0.5798085,"width":0.034242023,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Julia from Planhat","depth":17,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.07247341,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Julia","depth":18,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.00930851,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"from Planhat","depth":19,"bounds":{"left":0.91107047,"top":0.6927374,"width":0.026761968,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close","depth":15,"bounds":{"left":0.96758646,"top":0.6671987,"width":0.023936171,"height":0.057462092},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"Improved Task dependencies in Workflows","depth":17,"bounds":{"left":0.88613695,"top":0.7246608,"width":0.09707447,"height":0.0490822},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Improved Task dependencies in Workflows","depth":18,"bounds":{"left":0.88613695,"top":0.7258579,"width":0.09142287,"height":0.038707104},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:","depth":18,"bounds":{"left":0.88613695,"top":0.7869114,"width":0.094082445,"height":0.060255386},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Tasks can have multiple dependencies, across phases","depth":19,"bounds":{"left":0.8937833,"top":0.85714287,"width":0.08377659,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent task dates automatically update when a parent task is moved","depth":19,"bounds":{"left":0.8937833,"top":0.8910614,"width":0.07762633,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent tasks are assigned dates when a workflow is applied","depth":19,"bounds":{"left":0.8937833,"top":0.92498004,"width":0.07829122,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read more in our","depth":18,"bounds":{"left":0.88613695,"top":1.0,"width":0.037898935,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"docs opens in new tab","depth":18,"bounds":{"left":0.9240359,"top":1.0,"width":0.010472074,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"docs","depth":19,"bounds":{"left":0.9240359,"top":1.0,"width":0.010472074,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":18,"bounds":{"left":0.93450797,"top":1.0,"width":0.0013297872,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"","depth":18,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-1272291732256829089
|
-2831700515846899050
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab
docs
.
...
|
22653
|
NULL
|
NULL
|
NULL
|
|
22655
|
974
|
32
|
2026-05-12T07:17:08.126456+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570228126_m1.jpg...
|
Firefox
|
Content Explorer — Work
|
1
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.13090278,"top":0.0,"width":0.24027778,"height":0.026666667},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.13090278,"top":0.0,"width":0.18402778,"height":0.022777777},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-2318075514972954299
|
-2831770885395858542
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup...
|
22649
|
NULL
|
NULL
|
NULL
|
|
22656
|
975
|
42
|
2026-05-12T07:17:08.126461+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570228126_m2.jpg...
|
Firefox
|
Content Explorer — Work
|
1
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.02925532,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"L","depth":11,"bounds":{"left":0.9730718,"top":0.06304868,"width":0.002493351,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Lukas","depth":10,"bounds":{"left":0.9808843,"top":0.06304868,"width":0.012300532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.09377494,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.09936153,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Data Explorer","depth":12,"bounds":{"left":0.46426198,"top":0.11931365,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Data Explorer","depth":14,"bounds":{"left":0.47623006,"top":0.12490024,"width":0.027925532,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Calendar","depth":12,"bounds":{"left":0.46426198,"top":0.14485236,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Calendar","depth":14,"bounds":{"left":0.47623006,"top":0.15043895,"width":0.01861702,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Notifications","depth":12,"bounds":{"left":0.46426198,"top":0.17039107,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Notifications","depth":14,"bounds":{"left":0.47623006,"top":0.17597765,"width":0.026263298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"More","depth":12,"bounds":{"left":0.46426198,"top":0.19592977,"width":0.07114362,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"More","depth":14,"bounds":{"left":0.47623006,"top":0.20151636,"width":0.010638298,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Sections","depth":13,"bounds":{"left":0.46426198,"top":0.23264167,"width":0.07114362,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":15,"bounds":{"left":0.46559176,"top":0.23782921,"width":0.01662234,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"CS Day-to-day","depth":16,"bounds":{"left":0.47822472,"top":0.26456505,"width":0.030751329,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚀 Getting started Guide","depth":18,"bounds":{"left":0.47822472,"top":0.29010376,"width":0.049700797,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🪬 Just CS Data","depth":18,"bounds":{"left":0.47822472,"top":0.3140463,"width":0.032081116,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"👉 Daily Operations","depth":18,"bounds":{"left":0.47822472,"top":0.33798882,"width":0.03956117,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🗓️ Weekly prep","depth":18,"bounds":{"left":0.47822472,"top":0.36193135,"width":0.03125,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🤑 Renewals and Upsell","depth":18,"bounds":{"left":0.47822472,"top":0.3858739,"width":0.048204787,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"🚨 Risk and Churn Analytics","depth":18,"bounds":{"left":0.47822472,"top":0.40981644,"width":0.05718085,"height":0.01396648},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Implementation","depth":16,"bounds":{"left":0.47822472,"top":0.44972068,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Impl Projects","depth":18,"bounds":{"left":0.47822472,"top":0.47525936,"width":0.026761968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Trial Opps (Under Review)","depth":18,"bounds":{"left":0.47822472,"top":0.49920192,"width":0.054022606,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Stoyan's clients","depth":18,"bounds":{"left":0.47822472,"top":0.5231444,"width":0.032579787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Content Explorer","depth":15,"bounds":{"left":0.54355055,"top":0.09537111,"width":0.046043884,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":17,"bounds":{"left":0.55285907,"top":0.0981644,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon For you","depth":16,"bounds":{"left":0.5944149,"top":0.08739027,"width":0.028091755,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"For you","depth":18,"bounds":{"left":0.6043883,"top":0.0981644,"width":0.015458777,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Sections","depth":16,"bounds":{"left":0.62317157,"top":0.08739027,"width":0.030585106,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Sections","depth":18,"bounds":{"left":0.633145,"top":0.0981644,"width":0.017952127,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Pages","depth":16,"bounds":{"left":0.65442157,"top":0.08739027,"width":0.025265958,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pages","depth":18,"bounds":{"left":0.664395,"top":0.0981644,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Tab icon Archived","depth":16,"bounds":{"left":0.6803524,"top":0.08739027,"width":0.031083776,"height":0.035115723},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Archived","depth":18,"bounds":{"left":0.6903258,"top":0.0981644,"width":0.018450798,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":14,"bounds":{"left":0.84890294,"top":0.09377494,"width":0.036402926,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":16,"bounds":{"left":0.85837764,"top":0.0981644,"width":0.018118352,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"Search...","depth":16,"bounds":{"left":0.55502,"top":0.12809257,"width":0.05086436,"height":0.023942538},"on_screen":true,"help_text":"","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"No content to display","depth":19,"bounds":{"left":0.6924867,"top":0.56264967,"width":0.043882977,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace","depth":19,"bounds":{"left":0.6549202,"top":0.5802075,"width":0.119015954,"height":0.027533919},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Add new","depth":18,"bounds":{"left":0.6961436,"top":0.6209098,"width":0.03656915,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Add new","depth":20,"bounds":{"left":0.7056183,"top":0.6256983,"width":0.018284574,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Content Explorer","depth":18,"bounds":{"left":0.56981385,"top":0.91660017,"width":0.050033245,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Content Explorer","depth":20,"bounds":{"left":0.58111703,"top":0.9261772,"width":0.03474069,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.62117684,"top":0.91660017,"width":0.118351065,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.63248,"top":0.9261772,"width":0.10305851,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Organizing content: Pages, Sections and Libraries","depth":18,"bounds":{"left":0.7408577,"top":0.91660017,"width":0.11818484,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Organizing content: Pages, Sections and Libraries","depth":20,"bounds":{"left":0.7521609,"top":0.9261772,"width":0.10289229,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Explore Keynote","depth":15,"bounds":{"left":0.9243683,"top":0.3292099,"width":0.0390625,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Last edited","depth":16,"bounds":{"left":0.89744014,"top":0.36871508,"width":0.021110373,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Active Renewal Opps","depth":18,"bounds":{"left":0.9054189,"top":0.3934557,"width":0.04637633,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2 minutes ago","depth":18,"bounds":{"left":0.9659242,"top":0.39385474,"width":0.02443484,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"All Existing Clients","depth":18,"bounds":{"left":0.9054189,"top":0.41660017,"width":0.03507314,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"15 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4169992,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"My Renewals next 2Qs","depth":18,"bounds":{"left":0.9054189,"top":0.43974462,"width":0.04338431,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.44014364,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"OD Tasks","depth":18,"bounds":{"left":0.9054189,"top":0.46288908,"width":0.018118352,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"17 minutes ago","depth":18,"bounds":{"left":0.9645944,"top":0.4632881,"width":0.025764627,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"First 30 days to adoption","depth":18,"bounds":{"left":0.9054189,"top":0.48603353,"width":0.04737367,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"20 minutes ago","depth":18,"bounds":{"left":0.96359706,"top":0.48643255,"width":0.026761968,"height":0.011173184},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Explore community templates","depth":16,"bounds":{"left":0.89744014,"top":0.5203512,"width":0.05651596,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Page templates","depth":18,"bounds":{"left":0.90807843,"top":0.5510774,"width":0.029421542,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Section templates","depth":18,"bounds":{"left":0.90807843,"top":0.5798085,"width":0.034242023,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Julia from Planhat","depth":17,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.07247341,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Julia","depth":18,"bounds":{"left":0.90176195,"top":0.6927374,"width":0.00930851,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"from Planhat","depth":19,"bounds":{"left":0.91107047,"top":0.6927374,"width":0.026761968,"height":0.012769354},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close","depth":15,"bounds":{"left":0.96758646,"top":0.6671987,"width":0.023936171,"height":0.057462092},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"Improved Task dependencies in Workflows","depth":17,"bounds":{"left":0.88613695,"top":0.7246608,"width":0.09707447,"height":0.0490822},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Improved Task dependencies in Workflows","depth":18,"bounds":{"left":0.88613695,"top":0.7258579,"width":0.09142287,"height":0.038707104},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:","depth":18,"bounds":{"left":0.88613695,"top":0.7869114,"width":0.094082445,"height":0.060255386},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Tasks can have multiple dependencies, across phases","depth":19,"bounds":{"left":0.8937833,"top":0.85714287,"width":0.08377659,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent task dates automatically update when a parent task is moved","depth":19,"bounds":{"left":0.8937833,"top":0.8910614,"width":0.07762633,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Dependent tasks are assigned dates when a workflow is applied","depth":19,"bounds":{"left":0.8937833,"top":0.92498004,"width":0.07829122,"height":0.028332002},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read more in our","depth":18,"bounds":{"left":0.88613695,"top":1.0,"width":0.037898935,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"docs opens in new tab","depth":18,"bounds":{"left":0.9240359,"top":1.0,"width":0.010472074,"height":-0.09976053},"on_screen":false,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
288263493835329509
|
-2831700515846899050
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K
L
Lukas
Content Explorer
Content Explorer
Data Explorer
Data Explorer
Calendar
Calendar
Notifications
Notifications
More
More
Sections
Sections
CS Day-to-day
🚀 Getting started Guide
🪬 Just CS Data
👉 Daily Operations
🗓️ Weekly prep
🤑 Renewals and Upsell
🚨 Risk and Churn Analytics
Implementation
Impl Projects
Trial Opps (Under Review)
Stoyan's clients
Content Explorer
Content Explorer
Tab icon For you
For you
Tab icon Sections
Sections
Tab icon Pages
Pages
Tab icon Archived
Archived
Add new
Add new
Search...
No content to display
The content explorer contains the Pages and Sections you have access to, and lets you customize your workspace
Add new
Add new
Content Explorer
Content Explorer
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Organizing content: Pages, Sections and Libraries
Explore Keynote
Last edited
All Active Renewal Opps
2 minutes ago
All Existing Clients
15 minutes ago
My Renewals next 2Qs
17 minutes ago
OD Tasks
17 minutes ago
First 30 days to adoption
20 minutes ago
Explore community templates
Page templates
Section templates
Julia from Planhat
Julia
from Planhat
Close
Improved Task dependencies in Workflows
Improved Task dependencies in Workflows
Introducing highly requested features for task dependencies within Workflows. These allow project management to better correspond with real-world delivery:
Tasks can have multiple dependencies, across phases
Dependent task dates automatically update when a parent task is moved
Dependent tasks are assigned dates when a workflow is applied
Read more in our
docs opens in new tab...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22657
|
974
|
33
|
2026-05-12T07:17:09.231626+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570229231_m1.jpg...
|
Firefox
|
Content Explorer — Work
|
1
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-7033255905884690172
|
-2831770885664293998
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22658
|
975
|
43
|
2026-05-12T07:17:09.237592+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570229237_m2.jpg...
|
Firefox
|
Content Explorer — Work
|
1
|
ws.planhat.com/jiminny/home/content-explorer
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Content Explorer","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Content Explorer","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.02925532,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"J J Jiminny","depth":8,"bounds":{"left":0.46409574,"top":0.058260176,"width":0.03474069,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"J","depth":11,"bounds":{"left":0.46758643,"top":0.07940942,"width":0.0023271276,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Jiminny","depth":9,"bounds":{"left":0.47539893,"top":0.06304868,"width":0.016123671,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Search Jiminny ⌘K","depth":9,"bounds":{"left":0.6296542,"top":0.058260176,"width":0.21143617,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Search Jiminny","depth":11,"bounds":{"left":0.6402925,"top":0.06304868,"width":0.031416222,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"⌘","depth":10,"bounds":{"left":0.82629657,"top":0.06344773,"width":0.0039893617,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"K","depth":10,"bounds":{"left":0.8351064,"top":0.06344773,"width":0.0026595744,"height":0.011971269},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
4159630415606179326
|
-2831911623153173866
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Content Explorer
Content Explorer
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
J J Jiminny
J
Jiminny
Search Jiminny ⌘K
Search Jiminny
⌘
K...
|
22656
|
NULL
|
NULL
|
NULL
|
|
22659
|
975
|
44
|
2026-05-12T07:17:12.893860+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570232893_m2.jpg...
|
Firefox
|
Login — Work
|
1
|
ws.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Welcome back","depth":9,"bounds":{"left":0.48603722,"top":0.30526736,"width":0.14494681,"height":0.03830806},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Welcome back","depth":10,"bounds":{"left":0.48603722,"top":0.30925778,"width":0.070644945,"height":0.030726258},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with Google","depth":9,"bounds":{"left":0.48603722,"top":0.3754988,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with Google","depth":11,"bounds":{"left":0.5398936,"top":0.38028732,"width":0.04454787,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Continue with SAML SSO","depth":9,"bounds":{"left":0.48603722,"top":0.40422985,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Continue with SAML SSO","depth":11,"bounds":{"left":0.5362367,"top":0.40901837,"width":0.051861703,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"or","depth":10,"bounds":{"left":0.55634975,"top":0.44014364,"width":0.0043218085,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Email","depth":10,"bounds":{"left":0.48603722,"top":0.46727854,"width":0.011136968,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"you@example.com","depth":10,"bounds":{"left":0.49019283,"top":0.48643255,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"you@example.com","role_description":"text field","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Password","depth":10,"bounds":{"left":0.48603722,"top":0.5215483,"width":0.020279255,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Forgot password?","depth":10,"bounds":{"left":0.5940825,"top":0.5215483,"width":0.036901597,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXTextField","text":"••••••","depth":10,"bounds":{"left":0.49019283,"top":0.54070234,"width":0.13663563,"height":0.023942538},"on_screen":true,"help_text":"","placeholder":"Password","role_description":"secure text field","subrole":"AXSecureTextField","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Log in","depth":9,"bounds":{"left":0.48603722,"top":0.57821226,"width":0.14494681,"height":0.022346368},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Log in","depth":11,"bounds":{"left":0.5521942,"top":0.5830008,"width":0.012632979,"height":0.0131683955},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-493634218375676853
|
-2831913821102687342
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Welcome back
Welcome back
Continue with Google
Continue with Google
Continue with SAML SSO
Continue with SAML SSO
or
Email
[EMAIL]
Password
Forgot password?
[PASSWORD_DOTS]
Log in
Log in...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22660
|
974
|
34
|
2026-05-12T07:17:15.258375+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570235258_m1.jpg...
|
Firefox
|
Login — Work
|
1
|
NULL
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions....
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false}]...
|
5508579009360439157
|
-2867365299068372342
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions....
|
22657
|
NULL
|
NULL
|
NULL
|
|
22661
|
975
|
45
|
2026-05-12T07:17:15.258386+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570235258_m2.jpg...
|
Firefox
|
Login — Work
|
1
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.54988027,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.57701516,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
1481586658186100718
|
-2876372498323113334
|
click
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat...
|
22659
|
NULL
|
NULL
|
NULL
|
|
22662
|
975
|
46
|
2026-05-12T07:17:16.242606+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570236242_m2.jpg...
|
Firefox
|
Work — Mozilla Firefox
|
1
|
NULL
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es r Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
New Tab
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:...
|
[{"role":"AXStaticText","text& [{"role":"AXStaticText","text":"Login","depth":4,"bounds":{"left":0.32696143,"top":0.57222664,"width":0.009973404,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ws.planhat.com","depth":4,"bounds":{"left":0.32696143,"top":0.5830008,"width":0.027260639,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.074221864,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.08539505,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.10694334,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.11811652,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.1396648,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.15083799,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.17238627,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.18355946,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.20510775,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.21628092,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.23782921,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.2490024,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.27055067,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.28172386,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.30327216,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.31444532,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.33599362,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.3471668,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.36871508,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.37988827,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.40143654,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.41260973,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.43415803,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.44533122,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4668795,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.47805268,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.49960095,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.51077414,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.5323224,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5434956,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.56504387,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.57621706,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.57222664,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.5977654,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.6089386,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.6049481,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.632083,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.07741421,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.07741421,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.12609737,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.16999201,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.17238627,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.20590582,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.21548285,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.21628092,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.21707901,"width":0.06648936,"height":0.78292096},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.3320032,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.21548285,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.34317636,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.34557062,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.38707104,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.38906625,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.39026338,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4517159,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.45331204,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.47964883,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.47964883,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.50159615,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.50039905,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.612929,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.612929,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.612929,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.612929,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.63367915,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.63367915,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.6544294,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.6544294,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.7047087,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.7047087,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.726656,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.7254589,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.858739,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.858739,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.98244214,"width":0.11502659,"height":0.01755786},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9840383,"width":0.08809841,"height":0.015961707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.046210106,"height":-0.010375142},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.09507979,"height":-0.010375142},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
1761320631709701341
|
-2831770885664293998
|
visual_change
|
accessibility
|
NULL
|
Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es r Login
ws.planhat.com
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
Close tab
New Tab
New Tab
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22672
|
975
|
52
|
2026-05-12T07:17:31.365614+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570251365_m2.jpg...
|
Firefox
|
Work — Mozilla Firefox
|
1
|
app-eu3.planhat.com/login
|
monitor_2
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read app-eu3.planhat.com...
|
[{"role":"AXRadioButton","text [{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.0518755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.06304868,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.08459697,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.09577015,"width":0.16888298,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.11731844,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.12849163,"width":0.16140293,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.15003991,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.16121309,"width":0.18816489,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.18276137,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.19393456,"width":0.039228722,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.21548285,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.22665602,"width":0.04537899,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.2482043,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.25937748,"width":0.1200133,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.28092578,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.29209897,"width":0.19331782,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.31364724,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.32482043,"width":0.07646277,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3463687,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.3575419,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"bounds":{"left":0.24335106,"top":0.3790902,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"bounds":{"left":0.25664893,"top":0.39026338,"width":0.40475398,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"bounds":{"left":0.24335106,"top":0.41181165,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"bounds":{"left":0.25664893,"top":0.42298484,"width":0.10106383,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"bounds":{"left":0.24335106,"top":0.4445331,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"bounds":{"left":0.25664893,"top":0.4557063,"width":0.15159574,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"bounds":{"left":0.24335106,"top":0.4772546,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"bounds":{"left":0.25664893,"top":0.4884278,"width":0.1143617,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"bounds":{"left":0.24335106,"top":0.509976,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"bounds":{"left":0.25664893,"top":0.5211492,"width":0.012965426,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"bounds":{"left":0.24335106,"top":0.54269755,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"bounds":{"left":0.25664893,"top":0.55387074,"width":0.009474734,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"bounds":{"left":0.24335106,"top":0.575419,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"bounds":{"left":0.25664893,"top":0.5865922,"width":0.014960106,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3","depth":4,"bounds":{"left":0.24335106,"top":0.60814047,"width":0.07962101,"height":0.032721467},"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3","depth":5,"bounds":{"left":0.25664893,"top":0.61931366,"width":0.25681517,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"bounds":{"left":0.31067154,"top":0.61532325,"width":0.007978723,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"bounds":{"left":0.24617687,"top":0.6424581,"width":0.07413564,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.24617687,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.2571476,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.2682846,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.27942154,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.29055852,"top":0.97007185,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"bounds":{"left":0.42802528,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"bounds":{"left":0.43999335,"top":0.055067837,"width":0.010638298,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"WORK, Google Account: lukas.kovalik@jiminny.com","depth":12,"bounds":{"left":0.43733376,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Main menu","depth":12,"bounds":{"left":0.32696143,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Chat","depth":12,"bounds":{"left":0.40940824,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Open menu for conversation actions.","depth":12,"bounds":{"left":0.42270613,"top":0.103751,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXHeading","text":"Conversation with Gemini","depth":15,"bounds":{"left":0.3226396,"top":0.14764565,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Conversation with Gemini","depth":16,"bounds":{"left":0.3226396,"top":0.15003991,"width":0.1200133,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Copy prompt","depth":21,"bounds":{"left":0.34557846,"top":0.18355946,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":21,"bounds":{"left":0.36552528,"top":0.19313647,"width":0.06648936,"height":0.09577015},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"bounds":{"left":0.3226396,"top":0.19393456,"width":0.019946808,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\\n\\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.","depth":23,"bounds":{"left":0.36552528,"top":0.19473264,"width":0.06648936,"height":0.80526733},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.","depth":23,"bounds":{"left":0.36552528,"top":0.30965683,"width":0.06582447,"height":0.15043895},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Expand","depth":21,"bounds":{"left":0.4320146,"top":0.19313647,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Listen","depth":22,"bounds":{"left":0.43733376,"top":0.32083002,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"bounds":{"left":0.34624335,"top":0.32322428,"width":0.030917553,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"bounds":{"left":0.34391624,"top":0.36472467,"width":0.0003324468,"height":0.0007980846},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"bounds":{"left":0.34391624,"top":0.36671987,"width":0.04105718,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a summary of the changes in the provided code diff:","depth":24,"bounds":{"left":0.33294547,"top":0.367917,"width":0.0987367,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"AI Call Scoring Additions","depth":23,"bounds":{"left":0.33294547,"top":0.4293695,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Call Scoring Additions","depth":24,"bounds":{"left":0.33294547,"top":0.4309657,"width":0.063663565,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed Scores:","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.058011968,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Added strict instructions to treat the","depth":26,"bounds":{"left":0.34557846,"top":0.45730248,"width":0.09042553,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"bounds":{"left":0.40591756,"top":0.4792498,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.","depth":26,"bounds":{"left":0.34557846,"top":0.47805268,"width":0.09990027,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Breakdown:","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.052526597,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"*","depth":26,"bounds":{"left":0.39810506,"top":0.5905826,"width":0.0051529254,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Short Context:","depth":26,"bounds":{"left":0.40325797,"top":0.5905826,"width":0.037898935,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Now displays the","depth":26,"bounds":{"left":0.34557846,"top":0.5905826,"width":0.09690824,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Scorecard","depth":26,"bounds":{"left":0.3877992,"top":0.6113328,"width":0.03324468,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"name and the overall","depth":26,"bounds":{"left":0.34557846,"top":0.6113328,"width":0.10206117,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"AI Score","depth":26,"bounds":{"left":0.3721742,"top":0.632083,"width":0.02144282,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"(the average of per-rule scores).","depth":26,"bounds":{"left":0.34557846,"top":0.632083,"width":0.09857048,"height":0.037110932},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Full Context:","depth":28,"bounds":{"left":0.3565492,"top":0.6823623,"width":0.032579787,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Details the","depth":28,"bounds":{"left":0.38912898,"top":0.6823623,"width":0.028424202,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":29,"bounds":{"left":0.35854387,"top":0.70430964,"width":0.036236703,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).","depth":28,"bounds":{"left":0.3565492,"top":0.70311254,"width":0.09142287,"height":0.12011173},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Application Rules:","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.046708778,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.","depth":26,"bounds":{"left":0.34557846,"top":0.83639264,"width":0.10239362,"height":0.09936153},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Grammar and Formatting Cleanup","depth":23,"bounds":{"left":0.33294547,"top":0.96009576,"width":0.11502659,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Grammar and Formatting Cleanup","depth":24,"bounds":{"left":0.33294547,"top":0.9616919,"width":0.08809841,"height":0.016360734},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Typo Corrections:","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.046210106,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Fixed minor grammatical errors, such as changing \"contexts\" to \"context\" and \"utlizes\" to \"utilizes\".","depth":26,"bounds":{"left":0.34557846,"top":0.9880287,"width":0.09507979,"height":0.011971295},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"String Formatting:","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.04720745,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (","depth":26,"bounds":{"left":0.34557846,"top":1.0,"width":0.0965758,"height":-0.07980847},"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\\n\"","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") from the text strings.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":22,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Share & export","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":22,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy prompt","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Edit","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXHeading","text":"You said are there some issues and bottlenecks","depth":21,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"You said","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"are there some issues and bottlenecks","depth":23,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Listen","depth":22,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Show thinking","depth":25,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"Gemini said","depth":20,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Gemini said","depth":21,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Here is a breakdown of the potential issues and bottlenecks based on the changes:","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1. Token Consumption and Context Bloat (Bottleneck)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Massive Payload Size:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"By injecting the full","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"ai_call_score","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"per rule","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"), the token count per call will skyrocket.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Context Window Limits:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt states:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"The full call context includes all calls data available...\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"If a user filters for 5 or 10 calls, injecting the transcripts","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"plus","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the \"lost in the middle\" effect).","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"2. LLM Instruction Conflicts (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"2. LLM Instruction Conflicts (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Pre-computed vs. Raw Transcript:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt explicitly instructs the model:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Do not","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"invent, override, or recalculate scores\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"but later says,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\"For","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what was actually said","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", still rely on the","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"transcript","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".\"","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"justification","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":", the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Prompt Complexity:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The more rules you add about","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"how","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"to read JSON formatting and","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"what","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXHeading","text":"3. Edge Cases in Formatting (Issue)","depth":23,"on_screen":false,"help_text":"","role_description":"heading","subrole":"AXUnknown"},{"role":"AXStaticText","text":"3. Edge Cases in Formatting (Issue)","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Timestamp Limitations:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The instructions strictly define timestamps as","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":". This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"85:30","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":") or breaks if the system outputs","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"HH:MM:SS","depth":27,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":".","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Decimal vs. Whole Numbers:","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a \"Whole number","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"1–5","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"\". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.","depth":26,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?","depth":24,"on_screen":false,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Good response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Bad response","depth":21,"on_screen":false,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Redo","depth":21,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Share & export","depth":20,"on_screen":false,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Copy","depth":21,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Show more options","depth":20,"on_screen":false,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXTextArea","text":"Enter a prompt for Gemini\nencrypted","depth":20,"bounds":{"left":0.33626994,"top":0.8216281,"width":0.10638298,"height":0.01915403},"on_screen":true,"value":"Enter a prompt for Gemini\nencrypted","help_text":"","role_description":"text entry area","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Enter a prompt for Gemini","depth":21,"bounds":{"left":0.34291887,"top":0.82202715,"width":0.069980055,"height":0.018355945},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"encrypted","depth":21,"bounds":{"left":0.3352726,"top":0.8216281,"width":0.0066489363,"height":0.01915403},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Open upload file menu","depth":20,"bounds":{"left":0.33228058,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Tools","depth":18,"bounds":{"left":0.34823802,"top":0.8575419,"width":0.013297873,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Open mode picker","depth":20,"bounds":{"left":0.40525267,"top":0.85514766,"width":0.026097074,"height":0.031923383},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pro","depth":23,"bounds":{"left":0.4105718,"top":0.8639266,"width":0.007480053,"height":0.014764565},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXCheckBox","text":"Microphone","depth":19,"bounds":{"left":0.43334442,"top":0.85514766,"width":0.013297873,"height":0.031923383},"on_screen":true,"role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"Send message","depth":19,"bounds":{"left":0.4396609,"top":0.85434955,"width":0.013962766,"height":0.033519555},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":false,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.","depth":17,"bounds":{"left":0.32845744,"top":0.90901834,"width":0.11951463,"height":0.025139665},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXLink","text":"Your privacy & Gemini Opens in a new window","depth":17,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"link","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Your privacy & Gemini","depth":18,"bounds":{"left":0.41040558,"top":0.92178774,"width":0.040059842,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Opens in a new window","depth":19,"bounds":{"left":0.3226396,"top":0.92098963,"width":0.043218084,"height":0.012370312},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Summarize page","depth":7,"bounds":{"left":0.32829124,"top":0.95730245,"width":0.053523935,"height":0.025538707},"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Summarize page","depth":9,"bounds":{"left":0.33394283,"top":0.96249,"width":0.042220745,"height":0.015163607},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"Read app-eu3.planhat.com","depth":5,"bounds":{"left":0.46010637,"top":0.9876297,"width":0.047041222,"height":0.010774142},"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"}]...
|
-1740739471292033851
|
-2831913822176429422
|
visual_change
|
accessibility
|
NULL
|
New Tab
New Tab
Jy 20820 es reindex stream model h New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
app-eu3.planhat.com/?3kwcdhnccsyxrcjb4kjatxo2ay=uZzm4_8sOGXNUNik3cIZ3UFvxBAgd4w2-vQTwb1KM9wuTVpr_5nANjwoo3lxFh0Gb_20MAS3
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close
WORK, Google Account: [EMAIL]
Main menu
New Chat
Open menu for conversation actions.
Conversation with Gemini
Conversation with Gemini
Copy prompt
You said I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected. Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
You said
I’m on page “<tabTitle>JY-20361: Add call scores in Panorama by steliyan-</tabTitle>” with “<selection>@@ -4,7 +4,7 @@ Today is {date_today}.445System instructions:5System instructions:6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.6- You must answer using markdown. Do not use html tags in your response even if requested by the user's question.7-- The contexts includes data for call shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.7+- The context includes data for calls shortlisted through user-applied filters. Note that the calls might not relate to the same deal or not even be by the same sales rep.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.8 - The full call context includes all calls data available while the short call context contains only an overview of the calls that are analysed. Refer to the short call context for quick reference and to see the overall picture.9- The calls are ordered in chronological order. 9- The calls are ordered in chronological order. 10- Here’s how to use the call context:10- Here’s how to use the call context:@@ -16,6 +16,15 @@ System instructions:16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.16 - Frame answers with awareness of the company’s ICP, deal cycle, and sales motion.17 - Evaluate statements or objections based on how the team operates and what success looks like.17 - Evaluate statements or objections based on how the team operates and what success looks like.18 - Position responses in light of known competitors and market dynamics.18 - Position responses in light of known competitors and market dynamics.19+ - AI call score (`ai_call_score` in the full call JSON): When present, treat it as **pre-computed** output from your team’s AI call-scoring pipeline (the same kind of scoring as the `/call/ai-call-scoring` endpoint). **Do not** invent, override, or recalculate scores; interpret and summarize what is given.20+ - **Short call context:** **AI Scorecard** is the name of the scorecard applied to that call; **AI Score** is the **overall** score for that scorecard (the average of its per-rule scores, possibly shown as a decimal).21+ - **Full call context:** The `ai_call_score` object may include `ai_scorecard_name`, `score` (overall for that scorecard), and `ai_scorecard_rules`. For each rule when listed:22+ - `rule_name`: Title of the criterion.23+ - `rule_prompt`: The criterion text that was evaluated.24+ - `score`: Whole number **1–5** measuring how well the call satisfied that criterion against `rule_prompt` (1 = no evidence / not discussed or contrary; 5 = strong, clear evidence on the call).25+ - `justification`: Short rationale grounded in what happened on the call.26+ - `justification_timestamps`: Up to three entries with speaker **name** and **timestamp** (MM:SS in the recording) highlighting where the justification is supported.27+ - Use scores for coaching summaries, trends across calls, or quick comparisons when relevant. For **what was actually said**, still rely on the **transcript** (you may cross-reference rule timestamps when helpful).19- Use the Message History to:28- Use the Message History to:20 - Maintain continuity if the current question refers to previous exchanges.29 - Maintain continuity if the current question refers to previous exchanges.21 - Clarify or resolve ambiguities if the question depends on prior messages.30 - Clarify or resolve ambiguities if the question depends on prior messages.@@ -28,10 +37,10 @@ System instructions:28 - All factual claims must be supported by one or more references to relevant calls.37 - All factual claims must be supported by one or more references to relevant calls.29 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.38 - Use Markdown link syntax ([link text](URL)) and place links inline within sentences. Do not use footnotes, reference sections, or bare URLs. Integrate the reference links directly within the relevant parts of your response, rather than in a separate section.30 - Use markdown links in the format [link text](/playback/{{call_id}})39 - Use markdown links in the format [link text](/playback/{{call_id}})31- - Use descriptive link that utlizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.\n"40+ - Use descriptive link text that utilizes the call name. Do not use call time stamps in the link text even if the link itself contains a time stamp, e.g. use 'Call Name' instead of 'Call Name at 10:00'. Never use call ids as link text.32- - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)\n"41+ - Good example of link text: [Call Name](/playback/1234567890?apFrom=123)33- - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)\n"42+ - Bad example of link text: [Call Name at 12:34](/playback/1234567890?apFrom=123) (do not use call time stamps in the link text)34- - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)\n\n"43+ - Bad example of links: [Call Name on November 12, 2025](/playback/1234567890?apFrom=123) (do not use dates in the link text)35 - You might back your statements with examples from the provided call transcripts in addition to the reference links.44 - You might back your statements with examples from the provided call transcripts in addition to the reference links.36- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.45- Be specific. Use names of accounts, clients and persons involved. Don't just say "one client" or "a client"! If no account is available, use the call title as a reference.37- If the question or context is unclear, request clarification or highlight ambiguities.46- If the question or context is unclear, request clarification or highlight ambiguities.</selection>” selected.
Please summarize the selection using precise and concise language. Use headers and bulleted lists in the summary, to make it scannable. Maintain the meaning and factual accuracy.
Expand
Listen
Show thinking
Gemini said
Gemini said
Here is a summary of the changes in the provided code diff:
AI Call Scoring Additions
AI Call Scoring Additions
Pre-computed Scores:
Added strict instructions to treat the
ai_call_score
as pre-computed data. The system must interpret and summarize the provided scores rather than inventing or recalculating them.
Context Breakdown:
*
Short Context:
Now displays the
AI Scorecard
name and the overall
AI Score
(the average of per-rule scores).
Full Context:
Details the
ai_call_score
object, breaking down individual rules. Each rule includes a title, prompt, a 1–5 score, a brief justification, and up to three timestamped speaker references (MM:SS).
Application Rules:
Instructs the system to use these scores for coaching summaries, trends, and comparisons, but dictates that it must still rely on the actual transcript for direct quotes.
Grammar and Formatting Cleanup
Grammar and Formatting Cleanup
Typo Corrections:
Fixed minor grammatical errors, such as changing "contexts" to "context" and "utlizes" to "utilizes".
String Formatting:
Cleaned up the markdown link instructions by removing trailing newline and quotation artifacts (
\n"
) from the text strings.
Good response
Bad response
Share & export
Copy
Show more options
Copy prompt
Edit
You said are there some issues and bottlenecks
You said
are there some issues and bottlenecks
Listen
Show thinking
Gemini said
Gemini said
Looking closely at the system instructions in the diff, there are definitely a few potential prompt engineering issues and bottlenecks introduced by these additions. When you ask a Large Language Model (LLM) to juggle large transcripts alongside complex JSON scoring objects, it creates specific friction points.
Here is a breakdown of the potential issues and bottlenecks based on the changes:
1. Token Consumption and Context Bloat (Bottleneck)
1. Token Consumption and Context Bloat (Bottleneck)
Massive Payload Size:
By injecting the full
ai_call_score
object into the context (which includes rule names, prompts, scores, text justifications, and up to three timestamps
per rule
), the token count per call will skyrocket.
Context Window Limits:
The prompt states:
"The full call context includes all calls data available..."
If a user filters for 5 or 10 calls, injecting the transcripts
plus
these extensive AI scoring arrays could easily hit the model's maximum context window, leading to truncated data or degraded recall (the "lost in the middle" effect).
2. LLM Instruction Conflicts (Issue)
2. LLM Instruction Conflicts (Issue)
Pre-computed vs. Raw Transcript:
The prompt explicitly instructs the model:
"
Do not
invent, override, or recalculate scores"
but later says,
"For
what was actually said
, still rely on the
transcript
."
This creates a cognitive conflict for the LLM. If the raw transcript contradicts the pre-computed
justification
, the LLM might struggle with which source of truth to prioritize, potentially leading to hallucinations or contradictory summaries.
Prompt Complexity:
The more rules you add about
how
to read JSON formatting and
what
not to do, the less attention the model pays to the actual user query. This is a common bottleneck in prompt engineering known as instruction bloat.
3. Edge Cases in Formatting (Issue)
3. Edge Cases in Formatting (Issue)
Timestamp Limitations:
The instructions strictly define timestamps as
MM:SS
. This works fine for standard calls, but for longer meetings (over an hour), it either forces an awkward format (e.g.,
85:30
) or breaks if the system outputs
HH:MM:SS
.
Decimal vs. Whole Numbers:
The prompt specifies that the overall AI Score might be a decimal (average), but the individual rule score is a "Whole number
1–5
". If the upstream pipeline accidentally passes a decimal for a rule score, the LLM might get confused based on these rigid instructions.
How are you currently handling pagination or chunking for these calls—are you limiting the number of calls that can be passed into the full context at once to avoid token limits?
Good response
Bad response
Redo
Share & export
Copy
Show more options
Enter a prompt for Gemini
encrypted
Enter a prompt for Gemini
encrypted
Open upload file menu
Tools
Open mode picker
Pro
Microphone
Send message
Your Jiminny chats aren’t used to improve our models. Gemini is AI. It can make mistakes, so double check it.
Your privacy & Gemini Opens in a new window
Your privacy & Gemini
Opens in a new window
Summarize page
Summarize page
Read app-eu3.planhat.com...
|
NULL
|
NULL
|
NULL
|
NULL
|
|
22673
|
974
|
39
|
2026-05-12T07:17:33.233293+00:00
|
/Users/lukas/.screenpipe/data/data/2026-05-12/1778 /Users/lukas/.screenpipe/data/data/2026-05-12/1778570253233_m1.jpg...
|
Firefox
|
Login — Work
|
1
|
app-eu3.planhat.com/login
|
monitor_1
|
NULL
|
NULL
|
NULL
|
NULL
|
New Tab
about:newtab
New Tab
New Tab
Jy 20820 es r New Tab
about:newtab
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
Close tab
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close...
|
[{"role":"AXStaticText","text& [{"role":"AXStaticText","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXStaticText","text":"about:newtab","depth":4,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pipelines - jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pipelines - jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Pull requests · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Pull requests · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20773] User Pilot not receiving events on report generated - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"[JY-20776] Automated report - sentry - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"[JY-20776] Automated report - sentry - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"TypeError: League\\Flysystem\\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"LLM pricing overview with the most actual prices - Google Search","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"LLM pricing overview with the most actual prices - Google Search","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Planhat","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Planhat","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXRadioButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXStaticText","text":"New Tab","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXRadioButton","text":"Login","depth":4,"on_screen":true,"help_text":"","role_description":"tab","subrole":"AXTabButton","is_enabled":true,"is_focused":false,"is_selected":true},{"role":"AXStaticText","text":"Login","depth":5,"on_screen":true,"help_text":"","role_description":"text","subrole":"AXUnknown"},{"role":"AXButton","text":"Close tab","depth":5,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"New Tab","depth":4,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Customize sidebar","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Close Google Gemini (⌃X)","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Tabs from other devices","depth":6,"bounds":{"left":0.0,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open history (⇧⌘H)","depth":6,"bounds":{"left":0.019097222,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXCheckBox","text":"Open bookmarks (⌘B)","depth":6,"bounds":{"left":0.04236111,"top":0.0,"width":0.022222223,"height":0.035555556},"on_screen":true,"help_text":"","role_description":"toggle button","subrole":"AXToggle","is_enabled":true,"is_focused":false,"is_selected":false},{"role":"AXButton","text":"AI Chat settings","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false,"is_expanded":false},{"role":"AXButton","text":"Close","depth":7,"on_screen":true,"help_text":"","role_description":"button","subrole":"AXUnknown","is_enabled":true,"is_focused":false,"is_selected":false}]...
|
646656717547085956
|
-2444025834583934326
|
click
|
accessibility
|
NULL
|
New Tab
about:newtab
New Tab
New Tab
Jy 20820 es r New Tab
about:newtab
New Tab
New Tab
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
Jy 20820 es reindex stream model hydration by Vasil-Jiminny · Pull Request #12059 · jiminny/app
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
[JY-20725] [HubSpot] Optimise CRM rematching on delete hubspot accounts/contacts - Jira
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
JY-20725 add HS rate limit handling on activities rematching by LakyLak · Pull Request #12066 · jiminny/app
Pipelines - jiminny/app
Pipelines - jiminny/app
Pull requests · jiminny/app
Pull requests · jiminny/app
[JY-20773] User Pilot not receiving events on report generated - Jira
[JY-20773] User Pilot not receiving events on report generated - Jira
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
JY-20773 fix user pilot tracking ofr automated report generated by LakyLak · Pull Request #12024 · jiminny/app
[JY-20776] Automated report - sentry - Jira
[JY-20776] Automated report - sentry - Jira
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
TypeError: League\Flysystem\Filesystem::has(): Argument #1 ($location) must be of type string, null given, called in /home/jiminny/vendor/laravel/framework/src/Illuminate/Filesystem/FilesystemAdapter.php on line 218 — jiminny — app
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
Platform Sprint 3 Q2 - Platform Team - Scrum Board - Jira
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
JY-20625 | JY-20742 | MCP POC by yalokin-jiminny · Pull Request #12036 · jiminny/app
LLM pricing overview with the most actual prices - Google Search
LLM pricing overview with the most actual prices - Google Search
Planhat
Planhat
Login
Login
New Tab
New Tab
Close tab
Login
Login
Close tab
New Tab
Customize sidebar
Close Google Gemini (⌃X)
Tabs from other devices
Open history (⇧⌘H)
Open bookmarks (⌘B)
AI Chat settings
Close...
|
NULL
|
NULL
|
NULL
|
NULL
|