AI co-pilot
How the AI cost cap works
300 requests per tenant per day by default. Resets at 00:00 UTC.
Every call into runChatTurn (quick-add, auto-fill, conflict resolver, MISSED resolver, chat widget) bumps a per-tenant daily counter. When the counter exceeds the cap, further calls throw with a clear error message until the next UTC day.
Default is 300/day. Override via the AI_DAILY_REQUEST_CAP env var on the server (set to 0 to disable).
Grand admins can monitor usage from /admin/ai-usage: per-tenant rollup with peak day and "days at ≥80% cap" so you can spot runaway tenants before a cap fires or an upgrade conversation.