AI Token Budgets
How DataAgent manages AI usage to prevent runaway costs.
AI enrichment and chat both consume LLM tokens. DataAgent applies per-organisation token budgets to keep usage predictable.
Budget windows
| Budget | Default limit | Resets |
|---|---|---|
| Daily | 200,000 output tokens | Every 24 hours at midnight UTC |
| Weekly | 1,000,000 output tokens | Every Monday at midnight UTC |
Both limits apply simultaneously. Usage is gated by whichever limit is reached first.
What consumes tokens
| Action | Approximate cost |
|---|---|
| AI enrichment per incident | ~500–2,000 output tokens |
| Chat message (short answer) | ~100–500 output tokens |
| Chat message (detailed analysis) | ~500–2,000 output tokens |
Token counts vary based on the complexity of the incident and the length of the AI response.
When the budget is exhausted
- New incidents are detected and displayed as normal, but AI enrichment is paused.
- Existing chat conversations are paused — sending a new message returns an error until the budget resets.
- The incident detail view shows a notice when enrichment is unavailable due to budget.
Adjusting limits
Default limits are set at the organisation level. Contact support to request a higher budget.