The setup guide and how to know your workspace is healthy
The bottom-left setup launcher, the settings overview status dots, and the AI trust verdict — read together, they tell you exactly where you are.
There are two questions every workspace owner asks in the first month: "what's left to set up?" and "is it working?" Ochre answers both without a dashboard you have to ask for.
The setup guide launcher
In the bottom-left of the app, a small circular icon shows your setup progress as a ring. It's quiet until you hover — then a tray expands with the remaining setup steps grouped by stage: connect a channel, invite teammates, train the AI on your KB, set a routing rule. Each step is a single click into the right settings page, with the relevant section already scrolled to.
The launcher disappears entirely once setup is complete. It does not nag, it does not gamify, and it does not show you a green ring you have to dismiss. We trust that you know when you're done.
The settings overview with status dots
Open /settings and you'll see the new overview page: every major area as a card with a status dot. Green for healthy, yellow for needs-attention, red for broken, grey for not-configured. Each dot is computed from a real check, not a checkbox:
- Channels — green if every connected channel has received a real message in the last 7 days
- AI agent — green if your trust verdict is "ready" or "trusted" (more on this below)
- Knowledge base — yellow if any published article has been edited but not re-indexed
- Routing — yellow if you have rules that haven't matched in 30 days (probably stale)
- Integrations — mirrors the deliveries health dot
- Team — yellow if any pending invite is older than 7 days
Click a yellow or red card and you land on the page that shows you why. We deliberately do not surface "everything is fine" — green is the absence of trouble, not a celebration of it.
The AI trust verdict
The /ai page has a verdict at the top: a single sentence that summarises whether you should trust the AI agent to autopilot. The colour comes from a small bundle of measurements: escalation rate, confidence distribution, KB coverage, and the last week of human edits to AI drafts.
- Trusted (green) — autopilot is on and the numbers say it should stay on.
- Ready (green-yellow) — the numbers support autopilot, but you haven't enabled it yet. One click flips it on.
- Calibrating (yellow) — too few conversations to be sure. Keep using draft mode for a week and check back.
- Hold (red) — escalation rate or edit distance is high enough that we don't recommend autopilot today. The verdict explains what's driving it.
What each colour means, generally
Across the app, we use colour the same way:
- Green — measured-healthy in the last meaningful window
- Yellow — needs your attention, but not customer-impacting
- Red — customer-impacting now, fix today
- Grey — not yet configured, no signal either way
If you only check one thing per week, check the settings overview. Two minutes, full picture.
Was this article helpful?