Credits & metering
The assistant runs on prepaid credits — how they're deducted, how to top up, and what happens at zero.
The assistant is metered with prepaid credits, separately from your subscription. Each plan includes an allowance; you top up when you want more headroom. Credits keep heavy AI usage predictable instead of metered after the fact.
What each plan includes, and what a top-up pack costs, is on the pricing page and on Settings → Billing — never in these docs, so there is nothing here to go stale
How credits are deducted
A credit is consumed per assistant interaction that calls the model. Read-only browsing of your own resources is cheap; longer reasoning, doc search, and multi-step plans cost more. The running balance is shown in the assistant panel and on the billing tab.
Prop
Type
Topping up
Credit top-ups are part of the main billing tab — there's no separate flow. Purchases go through the same Stripe checkout as your subscription and land on your balance right away.
Assistant credits were merged into billing — manage your subscription and your assistant credits in the same place. Top-ups are a one-time charge, not a plan change.
At zero balance
When the balance hits zero, the assistant pauses write and reasoning actions and prompts you to top up. Your flags, configs, and data are unaffected — only the in-dashboard assistant is gated on credits.
Measurement plans
When a change needs instrumentation it can't do itself, the assistant writes a measurement plan and files measure_plan tickets for an agent to implement.
Assistant use cases
Three worked chat flows — ramp a feature with an alert on it, generate a measurement plan, and run a read-mode audit of your flags.