Configuring
The editor has four tabs — Setup, Test, Runs, Versions:- Setup — write the instructions yourself, copy them from a skill, or describe your intent and let AI generate the prompt. Pick the tools the field may read, and optionally limit the audience to a segment (with a live count of matched customers).
- Test — run the draft on one customer and watch the AI’s steps, then see the value, its reasoning, and the tools used. A batch test dry-runs up to 50 customers with no writes.
- Runs — history of computations, plus Run all to recompute every targeted customer on demand.
- Versions — every publish saves a snapshot.
Refresh schedule
Daily, weekly (Mondays), every 15 days, monthly, or manual only. Scheduled runs fire at 9:00 AM in the field’s timezone.FAQ
Do AI field values overwrite anything my team typed?
Do AI field values overwrite anything my team typed?
AI fields are their own fields, computed by AI — they don’t write into your standard or synced fields.
What does publishing require?
What does publishing require?
Non-empty instructions and at least one tool. Until you publish, the field only runs in tests.
Can I see why the AI chose a value?
Can I see why the AI chose a value?
Yes — test runs show the value alongside the AI’s reasoning and which tools it read.
