AGENTPROOF 4.0 · FIVE-MINUTE PILOT

Change the policy. Watch the transaction change.

This is the fastest way to understand AgentProof. The exact same autonomous purchase can be admitted, limited, escalated or denied depending on the enterprise policy that governs the agent. Nothing here touches production.

Simulate the checkpoint

Policy max spend ($) Proposed purchase ($) Allowed counterparty Actual counterparty Human approval above ($)

Try $100 policy / $500 transaction, then raise the policy to $1,000 and run it again.

Decision

READY

No production effect. No payment. No permit.

Press “Resolve transaction”.

What production adds

Isolated organization + test/live environments.

Scoped API key whose raw value is shown once and never stored.

Versioned policy profile with action, spend, counterparty, evidence and approval rules.

Authenticated resolution proof recording the exact policy version used.

Live ALLOW decisions receive a short-lived one-time execution permit.

The permit is consumed at commit and authority revocation is checked again.

Usage and audit streams preserve who evaluated what under which policy.

Move to the 30-day Founding Production Pilot → · Enterprise API discovery · OpenAPI · Conformance Lab · Book a 20-minute pilot call

Launch your 24-hour test tenant

Test tenant ready

Your API key is shown once. Copy it now.

The integration pattern

const ap = new AgentProof({ signPayment })

await ap.guard(request, async () => {
  // Runs only after:
  // 1. policy resolution
  // 2. proof authentication
  // 3. one-time permit consumption
  // 4. commit-time revocation recheck
  return paymentRail.execute()
})

Your existing wallet, agent runtime and payment rail stay in place. AgentProof protects the crossing.