Every AI decision, signed.
Verifiable independently,
for years.

A certificate contains: a unique certificate_id, the schema version (1.1), the decision outcome (ALLOW, BLOCK, or MODIFY), an enforcement_detail block recording what matched (or null if nothing matched), a SHA-256 hash of the input payload, the UTC timestamp of issuance, and an HMAC-SHA256 signature over the canonical payload. The input payload itself is never stored inside the certificate — only its hash — so the certificate is safe to transmit and store without exposing the original content.

The verification endpoint (POST /api/tve/verify-attestation) is a live service. During the verification call, the service confirms the HMAC-SHA256 signature is intact, that the schema matches the declared version, and that the payload hash is internally consistent. You do not need access to the original model, the enforcement pipeline, or any internal EVE infrastructure to call this endpoint — only the certificate JSON. For archival verification beyond the platform's retention window, Enterprise plans support private deployment where you hold the signing key.

Certificate signatures do not expire — an HMAC-SHA256 signature over a static payload remains verifiable indefinitely as long as the signing key is available. Retrieval via the GET /api/tve/certificates/{cert_id} endpoint is available for 30 days on Starter, 12 months on Growth, and custom periods on Enterprise. Certificates are self-contained JSON, so you can export and archive them in your own storage system for as long as you require — the signature remains verifiable against the verification endpoint independently of the platform retention window.

The current version of Proof is integrated with the EVE enforcement pipeline — certificates record the outcome of EVE's governance gate. The governed-generate endpoint accepts a decision_input payload and runs it through the full enforcement stack before issuing the certificate. Integration with third-party AI systems — where EVE acts as a governance proxy for decisions made by an external model — is on the Enterprise roadmap. Contact sales if this is a requirement.

CoreGuard is the enforcement engine — it decides whether a request is ALLOWED, BLOCKED, or MODIFIED using 126 deterministic enforcement pillars. Proof is the certification layer — it wraps the CoreGuard decision in a signed, portable certificate. CoreGuard runs whether or not you use Proof. Proof is an optional layer for teams that need to produce independently verifiable records of what CoreGuard decided, for audit, compliance, or contractual purposes. You can use CoreGuard without Proof; you cannot use Proof without CoreGuard in the current architecture.

A SOC 2 Type II audit of the Proof infrastructure is currently in progress. The certificate issuance and verification pipeline is designed to satisfy the availability and security trust service criteria — specifically the CC6 and CC7 control families — but we have not yet received a report opinion. We will communicate the audit status and results to customers directly when available. If SOC 2 attestation is a blocking requirement for your procurement, please contact sales to discuss our current readiness documentation and expected timeline.