Integration patterns
Three reference recipes used in evaluations and whitepaper reproductions. Each pattern uses the same REST verification API and open-source Noir proving stack.
AI agent gateoffline QRMerkle auditOpenAPIagentic AIEU AI Act
Pattern 1: AI agent pre-tool gate
Trending requirement under EU AI Act and enterprise AI governance: block autonomous tool calls until policy returns signed allow or deny.
1. Agent runtime intercepts tool invocation.
2. Your backend calls POST /v1/verify or identity verify with scoped API key.
3. AffixIO returns verdict + proof_digest.
4. Runtime allows or denies tool execution; stores proof for audit.
Sandbox: identity verify panel. Paper: agentic AI governance. Sector: AI agents.
POST /v1/verify
Authorization: Bearer YOUR_KEY
{"policy":"agent_tool_spend","attributes":{"limit_gbp":500,"approved":true}}
Pattern 2: Offline QR ticket at gate
Scan-to-prove for venues and transport. Proof embedded at issue; gate verifies without live database.
1. Issue ticket with embedded proof (edge or API).
2. Scanner validates proof locally; enforces spent-proof.
3. Reconcile audit digests when online.
Sandbox: ticket edge flows. Paper: WP-037. Product: anti-scalping tickets.
Pattern 3: Programme eligibility with Merkle audit
Government and regulated programmes need FOI-defensible outcomes without copying citizen records to every vendor.
1. Case system sends minimised eligibility inputs.
2. AffixIO returns allow/deny and proof_digest.
3. Auditor fetches GET /v1/merkle/proof/{digest} or uses Merkle verifier.
Paper: WP-038. Sector: government.