The control plane for autonomous AI
The authority every consequential AI action answers to.
PROVEN.
It sits beneath the model and decides what an agent may do. Its signed receipts can be checked by your board, your auditor, and any regulator, without a login.
We build your software on ICA. It can prove what its AI did.
ALEETH builds the software you need with ICA as its foundation, from the first line. Every AI action it routes through ICA is checked before it runs, sealed in a signed receipt, and open to anyone who wants to check it. What we build carries the Powered by ICA seal.
Two verifiers. Real receipts. Copies that must fail.
Signed ICA receipts from production sit beside tampered copies of them. Two independent verifiers, one in Node and one in Python on a different crypto library, check every one offline against ICA's pinned keys. They agree on every exhibit and refuse every copy. Open the room and your own browser re-checks the receipts.
No login. Run the kit with your network off, and nothing asks you to trust an ALEETH server.
Shipped. Not a roadmap.
Five things the plane actually does.
Policy before the act. A signed receipt after. A kill switch in the path. Continuous rating. Proof anyone can re-run offline. That is the product. Everything else is a page.
-
The call is denied until the plane says otherwise.
Rail Guard sits in the path, not beside it. Identity, purpose, tool, and arguments resolve before anything runs. Allow is earned, one act at a time.
The control plane -
A live agent that crosses the line is stopped mid-action.
Quarantine latches until a person releases it. There is no timeout that talks the agent out of containment. The stop never waits on its seal, and a signed receipt records it.
Containment -
Signed receipts you can check offline.
Ed25519-signed. Append-only by database trigger. Hash-chained. Anchored through OpenTimestamps.
Verify a receipt -
Continuous rating, not an annual snapshot.
A commercial customer holds a live ICA certificate, re-rated as it runs. Any holder can prove theirs on the public verifier. The math is the authority.
Look up a credential -
Proof anyone can re-run offline.
The ICA Proof Room: real production receipts beside tampered copies, checked by two independent verifiers against pinned keys and re-checked in your browser. Download the kit and run it with your network off.
The Proof Room
Proof is a record
you can check yourself.
Any AI control plane can show you a decision and a dashboard. An auditor, a regulator, and a court ask a harder question: can I check it myself? With ICA, a receipt can be checked offline with the published verifier, against a copy of ICA's public keys you saved and confirmed once.
- A record you have to trust.
- Meaningful only inside their system.
- Signed through HashiCorp Vault Transit, so the private key stays out of the ICA service.
- Append-only, hash-chained, Merkle-anchored.
- Checkable offline against a saved public key.
The gateway's core invariants are specified in TLA+: deny-by-default, no authority for a revoked agent, tenant isolation, fail-closed, and single-use capabilities. The check covers every input of the model, not a sample of tests, and CI re-runs it when the specs or the gateway decision code change. The model does not set production's enforcement mode. TLA+ stays on the gateway.
Relationship Trust is a metadata score, not a formal proof. It scores the trust between two parties from communication metadata alone, never the content, on a 0 to 100 scale. Four safety properties are built into the scorer: spoofing has a hard ceiling, an untrusted signal cannot inflate trust, no single event can race the score up, and no black-box model sits in the path. Send a spoofed stream yourself and watch the ceiling hold.
A claim is a slide. A refusal you can trigger yourself is a demonstration. Six run live on /verify.
Doctrine
We do not propose principles.
We enforce them.
ALEETH writes the rules autonomous systems answer to, and it answers to them itself. The ICA control plane holds to the same constitution it enforces on the systems it governs. Article XI, Self-Governance, makes that boundary structural. The systems ICA governs cannot quietly rewrite the framework that governs them.
LAW II · THE ZERO LAW
The absence of architecture is itself an architecture.
INSTRUMENT
Used before the wave arrives, it functions as governance. Used after, it functions as forensics.
ARTICLE XI · SELF-GOVERNANCE
The system itself operates within the same constitution it enforces.
The product
One plane. Five doors.
01 · CONTROL PLANE
Beneath the model. In the path.
Intercept. Resolve. Decide. Seal. Halt, gate, or quarantine while the act is still preventable.
→02 · PROOF ROOM
Real receipts. Copies that must fail.
Signed production receipts beside tampered copies, checked by two independent verifiers against pinned keys. Re-check them in your browser, or download the kit and run it offline.
→03 · COVERAGE
Every jurisdiction. Every standard.
348 frameworks mapped, from NIST AI RMF and ISO 42001 to the EU AI Act, SOC 2, HIPAA, and enacted state laws. Sentinel Packs span US federal, state, and EU.
→04 · PROVENANCE
Credential states are written to an Ed25519-signed, append-only chain, anchored via OpenTimestamps to a public ledger ALEETH cannot rewrite. Look up any credential from a browser.
→05 · VERIFY
Check the record yourself.
Paste any ICA receipt hash and your browser asks ICA’s public verifier and transparency log. No login and no account needed.
→Industries
Built for the institutions that answer for AI.
24 regulated industries in ten families. Each page shows the record your reviewer asks for, what ICA hands them, and where to start.
AI Governance Readiness Check
Where does your AI governance actually stand?
Answer six questions. Get an indicative Control Readiness Score and the gap between policy and proof, on the spot, with no login. The paid ICA Assessment is separate. It starts from one consequential AI workflow and maps what each AI can see, change and charge.
Indicative Control Readiness Score
Projected with ICA in place:
Your widest gaps
This is an indicative score from your profile, not a tested result. It is not an ICA certificate. The paid ICA Assessment is separate. It starts from one consequential AI workflow, with interviews, documents and a map of what each AI can see, change and charge. It needs no system access and no guest or customer data.