Agent-008 — Agent GovernanceHATS — Cyber InsuranceH33-74 — Portable ProofPQ Verified — PQ ValidationCryptographic APIs
AIR
The Cryptographic Lifecycle
How It WorksWatch DemoThe MapPatent Portfolio
Architecture
H33-RootAgent-ZeroQ-SignReplayTruth
Cryptographic Services
UpstreamFHE-IQBFVCKKSTFHE
Open Standards
HATSHICSPQ-Verified (New)CRVOISQ-Key
Verification
Conformance SuiteVerification GuaranteesHATS SpecificationArchitectureProtocol Stability
Proof Lab
Cyber Insurance (HATS)BankingGovernmentHealthcareLaw FirmsPayments (WireProof)AI Companies
SDKs & Tools
SDKs HubTypeScript SDKMCP SDKVerifier CLIGitHub
References
DocumentationAPI ReferenceConformance VectorsSchemasBenchmarks
AboutPressPartnersPricingContactLegal
StartEcosystem
White PapersBenchmarksProofsResearch Overview
Explore (579)Live Systems (52)Pricing
Log InGet API Key✓ Verify It Yourself
H33 H33-74 ← All one-pagers Screen-shot the card · Cmd+Shift+4 · paste to LinkedIn
H33
H33-74Chain-Portable Evidence
For Chief Trust Officers

"Model provider change loses the audit trail for every prior decision the model made."

Every AI decision survives the model and the provider.

Related · tier-1 reading. For what a portable artifact actually is, see Portable Artifact.

01

EU AI Act Article 12 logging built in

Each inference, classification, and agent action emits a logging entry that satisfies Article 12 automatic logging. Article 14 oversight events and Article 17 quality management actions also produce proofs.

02

Agent action attestation per tool call

Every tool invocation, every decision an agent takes on a user's behalf, every escalation produces a verifiable proof. Agent activity becomes audit-grade.

03

Model-portable decision evidence

Change models, change providers. The audit trail of every prior decision stays verifiable independent of the model that produced it.

04

RLHF judgment provenance

Every preference judgment in human feedback emits a proof. The reward model's training history is auditable for years after the model has been replaced.

The asset may move.
The evidence remains.
h33.ai/h33-74
H33-74 · Post-Quantum Evidence

H33-74 in brief — for the Chief Trust Officer

H33-74 is the portable 74-byte post-quantum evidence primitive — a 32-byte on-chain commitment plus a 42-byte off-chain receipt, signed under three independent NIST post-quantum signature families (ML-DSA, FALCON, SLH-DSA) — that produces and anchors a self-contained proof of each AI decision, agent action, and oversight event. For a Chief Trust Officer it converts "trust us, here are our logs" into verifiable evidence that anyone can check — and that outlives the model, the provider, and the platform.

What it is not. H33-74 is a primitive, not a service. It is not continuous monitoring (that is HATS), not the verification verdict (that is Verification), and not agent governance (that is Agent-008). It records what happened as proof; it does not decide policy or intervene.

Why does a Chief Trust Officer care?

Responsible-AI and trust claims become defensible. Each inference, tool call, and preference judgment is a proof that survives a model or provider change, so the audit trail behind a public commitment does not disappear with the model.

How is it different from a logging platform?

Logs record what a system says happened; H33-74 produces cryptographic proof independent of that system. Trust rests in the bytes and an open verifier, not in the platform's own record.

What proves it?

Every claim resolves to a reproducible artifact — byte-exact wire format at /schemas/h33-74/ and independently reproducible timings at /benchmarks/.

Can a regulator verify without trusting us?

Yes. Any party runs an independent open verifier offline — recomputing the commitment, checking all three signatures, confirming the anchor — with no contact to H33 and no access to your systems.

Go deeper: What H33-74 is & how it works · How verification works · For General Counsel · For the CCO