Answer 6 questions about one AI agent. Get a plain-English spending policy, the exact rules JSON, and a live test against a free sandbox that shows the allow, ask and block decisions happen for real.
Updated 2026-10-05
Pink Agentic AI Payments enforces per-agent spending policies at the MCP layer, before a payment executes.
Payments at or under this amount go through with no human in the loop.
If this agent's category spend would pass this amount in one day, the payment is blocked outright, even for someone who could normally approve it.
A spending policy is a set of rules that decide, for every payment an AI agent tries to make, whether it is allowed automatically, sent to a human for approval, or blocked. Rules are checked in order and the first one that matches a payment decides the outcome.
Each answer maps to one field in a Pink rule: what the agent buys becomes a payee category, your auto-approve limit becomes an allow rule's maximum, your approver choice becomes that rule's approvers, and an optional daily hard stop becomes a block rule with no approval path. The generator never invents a rule shape the sandbox does not support.
No. Try it live creates a free sandbox workspace with test money and test payees, loads your generated rules into it, and fires a few policy checks so you can see the allow, ask and block decisions. No real payment ever executes.
The rules JSON is the format the Pink sandbox API accepts today, and the copyable curl script shows exactly which endpoints load it. Production access is by early access; if you want to carry a sandbox policy over, request early access on pinkwallet.com/agentic.
The default is block. The policy engine checks rules top to bottom and stops at the first match; if nothing matches, the payment is blocked, not allowed by default.