The card lifecycle
Six steps,
all of them auditable
Cards are short-lived by design. Each one exists for a single job, carries a single budget and closes the moment the job is done.
Your agent calls the API with the job, the merchant and a budget. No card details ever live in its prompt or its memory.
Your policies run before anything is issued: budget caps, merchant locks, category rules and velocity limits.
Above your threshold, a person gets the final say in Slack, email or SMS. Below it, the agent keeps working.
A single-use virtual card is created for that merchant and that amount, and handed to the agent as a scoped token.
The agent checks out like any customer would. The charge settles on the card network in real time.
The card closes after one payment or at expiry, whichever comes first. The receipt and the full trail stay in your logs.
Events
Every step,
as a webhook
Subscribe once and your systems hear about every card the moment something happens to it. Signed payloads, retries included.
Controls
Rules that run
before the charge
Budgets, locks and approvals are checked on every authorization, not reported afterwards. A card that breaks a rule is declined on the spot.
See pricingGet started
Give your agent
its first card
Issue a single-use card from code in a few minutes. The free plan covers 100 cards a month.
100 free cards on signup. No credit card needed.