The job of this console
Filter domain, open a card, follow the source, act there or mark the card unhelpful. Never copy unmasked pay or medical from a card into notes.
Analytics · AI
AI Insights
Insight cards composed from canonical entities — each card cites a source id, never a standalone AI ledger.
Scroll to Flowcharts. Scenario: Review analytics AI insight cards.
Hint cards
Every card owes a source id
AI insights are hint cards on Employee, Project, Task, PO, and similar sources. They are not a second system of record.
Analytics → AI insights
Route
/analytics/ai-insights
Org flag aiInsightsEnabled
Executive · Manager · Admin
Seat at the table
Executives, managers, admins with aiInsightsEnabled and scope. Others must not see cross-team cards.
The working rhythm
Scan cards, open one source, decide, feedback, next card.
Moments that bring you here
Monday pack. After a control-tower red. When an agent suggested something you want to verify.
ⓘ
Hints, not truth
Masked summaries only. Agent console is chat. This page is the card register on the same composers.
Flowcharts
1 — Card then the cited entity
Read a card, open its source, ignore it if the source disagrees.
flowchart TD
A[Open AI insights] --> B[Filter domain]
B --> C[Open a card]
C --> D[Source entity]
D --> E{Source agrees?}
E -->|No| F[Discard / feedback]
E -->|Yes| G[Act on source]
- 1OpenAnalytics → AI insights.
- 2FilterDomain chips.
- 3Open cardSee source type and id.
- 4Act on sourceOr mark not helpful.
Overview
AI insights are composed cards with sourceEntityType and sourceEntityId. They are not business truth. Sensitive fields stay masked. Agent console is conversation; this page is the card register.
How to use
1. Always open the source
A confident card can still be wrong.
2. Cross-team scope
Cards outside persona should not appear.
3. Audit is separate
AI decision audit stores masked run metadata.