AI behavior
The default panel in Assistant Settings at /settings/assistant (Sparkles icon). It controls how your Executive Assistant (EA) talks to you and what it does when it's out of its depth.
Controls
Three fields, stored as settings on your user record and applied on every response.
- Communication Style — dropdown with Concise (brief, direct), Balanced (default), and Detailed (comprehensive).
- Auto-escalate Complex Queries — toggle. On by default. Hands off to a human path when your EA detects it can't answer well; off leaves the EA to attempt anyway.
- Allow Assistant Learning — toggle. On by default. Lets your EA adapt to your tone, formality, and priorities between sessions.
Click Save Settings at the top right. The button shows "Saving…", then a three-second "Settings saved successfully" confirmation. Save posts PATCH /api/user/assistant/settings with the full { settings, workContext, privacy } payload — every section persists together. Changes take effect on the next message.
FAQ
Does "Concise" change length or just tone? Both. It removes preamble, hedging, and caveats; long answers still happen when a task needs them.
What does auto-escalate actually do? When the EA detects low confidence, the conversation hands off to a human path instead of producing a low-confidence answer. The escalation channel depends on your organization's configuration.
If I turn off Assistant Learning, do I lose what's already calibrated? No. It stops future learning; the current calibration is kept. Contact support for a reset.
For organization-wide assistant behaviour (Chiro), see Organization admin, not this page.