Gravitre AI (Chat mode) (/ai?mode=chat) is Gravitre's streaming chat interface for org-aware questions, troubleshooting, and light operational tasks. Legacy /assistant redirects here. It is not Gravitre AI (Execute mode) and not Universal Search.
Three AI surfaces (not one master chat)
| Surface | Route | Job |
|---|---|---|
| Gravitre AI | /ai?mode=execute | Delegate tracked work with plans and async jobs |
| Gravitre AI | /ai?mode=chat | Multi-turn conversation with tools and briefings |
| Universal Search | /search | Find records — returns links, not chat replies |
Chat mode vs Execute mode
| Gravitre AI (Chat mode) | Gravitre AI (Execute mode) | |
|---|---|---|
| Interaction | Multi-turn chat with history | Work console for plans and monitored jobs |
| Best for | Q&A, exploration, quick fixes | End-to-end work assignment and execution tracking |
| Route | /ai?mode=chat | /ai?mode=execute |
Both can use connected Connectors when configured; each run shows what was analyzed and what executed.
Key capabilities
- Conversation history — sidebar of past threads, resumable sessions
- Intelligence modes — Fast, Standard, Reasoning, and Agent modes for latency vs depth tradeoffs
- Model selection — Auto or explicit provider models (GPT, Claude, Gemini, etc.)
- Org context pill — surfaces what organizational context is in scope for the reply
- Tool chips — shows connector lookups and actions attempted during a turn
- Daily briefing — optional summary bullets linking to runs, workflows, or connectors that need attention
- Follow-up suggestions — suggested next questions after each reply
When to use Chat mode
- "Which connectors failed sync overnight?"
- "Summarize open approvals for my team"
- "Draft steps to fix a recurring workflow error"
For durable assigned work with explicit plans, prefer Gravitre AI (Execute mode). To locate a record by keyword, use Universal Search first.
Next steps
- Use Gravitre AI (Chat mode)
- Universal Search for cross-entity lookup without chat
- Use Gravitre AI (Execute mode)