A standard you can meet, and anyone can check.

Regulayer Class defines what a governed AI deployment looks like, modelled on the standards that already govern contamination control: ISO 14644 for semiconductor cleanrooms, and USP 797 for sterile compounding. Each class is a published verification floor.

Operators self-attest conformance against the standard. The attestation is signed, and any third party can verify it. Regulayer does not audit. The floor is public, the proof is portable, and the check belongs to whoever you hand it to.

Three classes. Each one a verifiable floor.

Class 1

Foundation

Behavioral drift is measured continuously, every governed decision produces a record built to support admissibility, and the provenance of every output is tracked from creation through delivery.

The entry floor a buyer can verify. For most enterprise AI deployments.

Class 2

Intermediate

Everything in Class 1, plus continuous operator attestation and runtime proof. The operator attests, the system measures itself, and both seal into the same decision.

For regulated industry, clinical decision support, financial automation, and legal AI inside firms.

Class 3

Cleanroom

The full kernel runs in architectural isolation. No data leaves the deployment, and the verifying body is not employed by the deployer. The highest floor.

For physical AI, healthcare, defence, export-controlled and sovereign deployments.

Same name. The school.

You do not just buy it. You learn to meet it.

A practitioner pathway built on the same three floors as the standard.

Foundation course

Self-paced

Eight modules, a recognised reading list, and a capstone deployment audit. Open enrollment.

Procurement leads, compliance officers, AI buyers, counsel evaluating vendors.

Intermediate cohort

Six weeks, live

Live sessions, a written exam, and a working deployment assessed against the Class 2 floor.

Technical architects, regulated-industry leads, data-protection officers, in-house counsel.

Cleanroom Architect

Twelve weeks, mentored

Cohort and mentor pairing, a full architecture review of a candidate deployment, and an independent assessor.

CTOs, heads of AI governance, sovereign and defence leads, certifying auditors.

How the classes relate to the standards you already run under.

Each class maps defined Regulayer evidence capabilities to relevant control and record-keeping concepts in established frameworks. The mapping does not establish conformance with the framework itself. A class identifies evidence capabilities that may support those control areas.

ISO/IEC 42001AI management system. Relevant evidence capabilities: Class 1 and up.

ISO/IEC 27001Information security across the AI surface. Relevant evidence capabilities: Class 2 and up.

ISO 14644Cleanroom classification, applied to architecture. Relevant evidence capabilities: Class 3.

USP 797 (2023)Contamination control. The founding principle of the kernel.

FDA 21 CFR Part 11Electronic records and signatures. Relevant evidence capabilities: Class 1.

EU GMP Annex 11Computerised systems in regulated environments. Relevant evidence capabilities: Class 1.

ISO/IEC 23894Risk management for high-risk AI. Relevant evidence capabilities: Class 2.

NIST AI RMFAccuracy, robustness, audit. Relevant evidence capabilities: Class 1.

The move.

When a powerful technology arrives, gatekeeping does not work.

So we do not restrict who can build governed AI. We define what governed AI looks like, and we teach it. The standard is published as a reference. Regulators, standards bodies, and governance institutions are welcome to read it, cite it, and build on it.

Define the floor. Teach it. Let anyone check.

Join the launch list Back to start

The cleanroom standard for AI.