Stateless verification

Stateless verification is a decision model: each transaction is evaluated from current context and returns a deterministic YES/NO outcome plus evidence. You don’t “trust the session.” You verify permission at the moment of transaction.

See a YES/NO decision.
Run the live demo, then map the model to consent receipts and replay safety.
Try the live verification demo Consent receipts

What stateless means in practice

No long-lived trust

Tokens can authenticate, but they are not sufficient proof of permission. Verification is transaction-scoped.

Auditable outputs

The verifier outputs a proof record that can be logged and referenced in disputes and audits.

Why this matters for AI agents

Verifier input/output (reference)

{
  "input": {
    "agent_id": "agt_123",
    "permission_ref": "cr_456",
    "context": { "action": "payment.authorise", "amount": 4200, "currency": "GBP", "merchant": "m_001" }
  },
  "output": {
    "eligible": true,
    "proof": "sha256:…",
    "policy_version": "pv_2026_03"
  }
}

How it connects to consent receipts and replay safety

Stateless verification needs a permission artifact (consent receipt) and replay constraints (nonce semantics) so that the same authorization can’t be reused. For offline/edge, apply bounded policies and reconcile on reconnect.

FAQ

Is stateless verification always cryptographic?

No, but the output should be evidence-bearing. Cryptographic binding helps when multiple parties need to trust the result (issuers/merchants/regulators).

Does this increase latency?

It can be tens of milliseconds when implemented as a stateless decision layer, and can run in parallel with existing checks.