Per-org cost guardrails
LiveA monthly AI token quota per plan, an 80% warning by email and in-app, and a hard stop at 100%.
- ✓Monthly token allowance per plan, with per-organisation overrides available
- ✓Cache reads counted at a tenth, so prompt caching stretches the allowance
- ✓One 80% warning per calendar month, by email and in-app, to owners and admins
- ✓Hard stop at 100% with an explicit response, not a silent failure
What it is.
Your plan carries a monthly AI token allowance — 500,000 on Starter, 2.5 million on Pro, 7 million on Business, 50 million on Enterprise. Usage counts input, output and cache-creation tokens in full and cache reads at a tenth, so caching genuinely stretches the allowance. Cross 80% and owners and admins get an email and an in-app notification, once per calendar month rather than every day. Reach 100% and further AI calls are refused with a clear payment-required response until the month rolls over or the allowance is raised.
Two corrections to what this page used to say. The allowance is a token quota, not a money ceiling — a daily spend cap is a separate, optional setting an admin can add in AI Administration, and the platform keeps its own daily cap on trial accounts with no plan, as runaway-spend protection. And there is no overage mode: 100% is a stop, not a meter.
Three steps.
End to end.
Nothing to configure. Per-organisation increases are possible by arrangement.
Email and in-app to owners and admins, once per calendar month.
Further AI calls are refused with a payment-required response naming the quota, until the month resets.
What you get.
- ✓Monthly token allowance per plan, with per-organisation overrides available
- ✓Cache reads counted at a tenth, so prompt caching stretches the allowance
- ✓One 80% warning per calendar month, by email and in-app, to owners and admins
- ✓Hard stop at 100% with an explicit response, not a silent failure
- ✓Race-safe reservation, so concurrent requests cannot both slip past the ceiling
- ✓A platform daily cap that applies to trial accounts without a plan
Quick answers.
Yes, since October 2026. An admin can set a daily AI spend cap, in your billing currency, in AI Administration; once today's spend reaches it, further AI calls are refused until the next day (UTC). Speech transcription is a separate path and is not covered by the cap. See AI Administration.
No. At 100% the answer is a stop. There is no pay-past-the-cap setting.
No — email and in-app only. Nothing in the product posts to Slack.
More in AI Controls & Guardrails.
Every citation comes back with a verdict — verified, ambiguous or unresolved — and a match on title alone is never verified.
Each legal claim we hold about a jurisdiction is checked against that government's own legislation service — and where it cannot be, the answer is "unverified", not silence.
We do not train models on customer content. There is no opt-in, because there is no training programme.
User-pasted text is sanitised before reaching the LLM.