Administration Govern and monitor
Usage, quotas and retention
Set monthly AI allowances, add group budgets and per-person overrides, and decide how long data is kept.
For Org Admins
Quotas
Every AI call is recorded: who, which model, how many tokens and the cost. You use that to set monthly limits.
Open Admin › Quotas & usage.
| Control | What it does |
|---|---|
| Organization default | A monthly limit per person, in tokens or cost. |
| Per-person override | A different limit for one person who needs more or less. |
| Group budget | A shared monthly budget that all the group's members draw from. |
What people experience
- At 80% (adjustable), the chat shows a warning.
- At 100%, new AI requests are refused until the next calendar month or until you raise the limit.
- Org Admins are exempt.
- Automations count against their owner. A task run through a Front Desk counts against the person who sent the request.
- A rate limit per person (requests per minute) protects against runaway loops.
Limits reset on the first of each calendar month. People can see their own usage in Settings › Usage.
What the usage report shows
A summary of AI use across the organization, from the same records the quotas use. Use it to decide where to give more, and where to look for waste.
Retention
How long data is kept is yours to set under Admin › Retention. The default is 180 days for each:
| Kind | Notes |
|---|---|
| Conversations | Automation conversations are excluded. Agents' shared run files follow this period. |
| Agent logs | The step-by-step records of what agents did. |
| Access logs | The audit trail. Keep this longer if regulation requires. |
| Automation runs | Run history. |
| AI usage records | Never deleted inside the current month, since quotas need them. |
The system purges daily at 03:00. Purge now runs it immediately, and each purge is itself written to the audit log.
Retention for AI content held in your observability system is configured there, not here.