AI agent payment verification demo

This page exists to make the category concrete: verification is a transaction-time YES/NO decision with a proof record you can store. Not a dashboard screenshot. Not a “we logged it.” Proof you can hand to an issuer, auditor, or merchant risk team.

Run the live demo.
Get a binary decision + proof object in seconds.
Open the live demo Consent receipt verifier (lab)
YES: in scope NO: out of scope proof: portable evidence

What the demo should show (acceptance criteria)

Transaction-scoped inputs

Verification is evaluated on the proposed transaction context (amount, merchant/category, time window), not a broad token.

Binary decision

The output is a crisp YES/NO signal. No ambiguous “risk score only” outcomes.

Proof object

The result produces an auditable record (hash/receipt id/policy version) you can attach to authorization and order records.

What inputs matter

Example decision record (what you want to log)

{
  "eligible": true,
  "agent_id": "agt_123",
  "permission_ref": "cr_456",
  "policy_version": "pv_2026_03",
  "proof": "sha256:…",
  "evaluated_at": "2026-03-18T12:00:00Z"
}

How to use this in real payment flows

Issuer path

Verify before authorization. Treat the decision as a first-class signal alongside fraud/standing checks.

Issuer AI agent authorization →

Merchant path

Verify before order creation / checkout acceptance. Bind agent identity to session/token and enforce policy.

Merchant AI agent verification →

Related pages

FAQ

Is the live demo the same as the lab consent receipt verifier?

No. The live demo is the decision model end-to-end; the lab ACRRV focuses on the consent receipt + replay protection layer.

What should I expect to integrate?

A pre-authorization or pre-checkout verification call, plus logging the proof output alongside your payment/order record.