Summary
A notification system that serves two purposes: operational alerts (artefact ready, task complete) and proactive intelligence alerts (critical findings in user data).
Notification Types
Operational Notifications
- Artefact ready — background generation has completed
- Background task complete — a suggested background task from exploration has finished
- Sharing activity — someone viewed/commented on your shared artefact
- Team activity — artefacts shared with you, team invitations, etc.
Intelligence Notifications (Proactive Alerts)
- Critical findings — issues discovered in the user's documents
- Compliance breaches
- Policy violations
- Data inconsistencies
- Risk indicators
- These could be surfaced by agents during ingestion, scheduled analysis, or exploration
- Higher urgency — should be visually distinct from operational notifications
UI Placement
- Notification icon/bell in the main navigation
- Badge showing unread count
- Dropdown or panel showing recent notifications
- Full notifications page for history and management
Notification Behaviour
- Real-time delivery (push or polling)
- Read/unread state
- Click-through to relevant content (e.g. click "artefact ready" → opens artefact)
- Dismissable
- Possibly grouped by type or time period
Open Questions
- Should notifications have email/push delivery as well as in-app?
- How are intelligence notifications generated? On ingestion? On schedule? On demand?
- Should users be able to configure notification preferences? (Which types, delivery method, frequency)
- How do we prevent notification fatigue?
- Should there be a notification centre with filters and search?
- Are notifications tied to the user or to the workspace/team?
- How long are notifications retained?
- Should critical/intelligence notifications have an "acknowledge" action (vs. just "read")?