When an agent does the work, the natural question is how much work got done, and that is what AI credits measure. Usage-based billing ties what you pay to the work your agent actually performs. Here is what a credit is, how the billing works, and how to keep spend predictable.
What an AI credit is
A credit is a unit of work your agent performs. You hold a balance, and as the agent acts, it draws down against it. The result is a direct line between what you spend and what you got: real money for real work, measured action by action.
What you are paying for
AI usage is billed as credits: every action your agent takes draws from a balance you hold, so what you spend tracks the work actually done. When things are quiet, usage falls and so does the cost. With an agent doing the work, spend maps cleanly to the value you get, and you can see exactly where it went.
Keeping spend predictable
Usage-based does not mean unpredictable. You set a budget and a cap, turn on auto-recharge so the balance refills without stalling your agents, and see exactly where every credit went, by action and by app. If you prefer steadier costs, a monthly plan bundles a generous allowance at a better rate, so you get the predictability of a subscription with the fairness of paying for use.
Free to start
Launching a workspace costs nothing. You pay as you go for the work your agent does, and you can top up whenever you like, with larger top-ups carrying a better rate. See the pricing page for the current figures.