What a licensed body receives
An annual platform licence with a co-branded method. It includes the protocol and its reference implementation, the hosted platform on which your assessors pre-register, run and seal evaluations, the held-back item store for one domain, assessor training and annual recertification, and a record corpus that grows with every evaluation you run and shows your consistency to your notifying authority. Ten evaluations a year in one domain are included; further evaluations and further domains are priced on top.
Every evaluation your assessors run produces the same artefacts described on the method page: a hash-committed pre-registration, a published harness, an append-only record with independently republished roots, and a machine-checked certificate that the procedure was followed as registered.
Rather than building one
A body that builds its own AI assessment method needs research scientists it does not employ, a formal verification team it has never needed, and eighteen months it does not have. A body that licenses ours has a method a technical assessor can read closely and a record it can put in front of a notifying authority on the day of designation. The licence is priced below what ten in-house assessments cost at current assessor day rates, and the corpus, the certificate and the neutral-operator position cannot be reproduced inside the window before Annex III applies.
The credit stays with the body. Antefacts is not an accredited certification body and does not issue conformity certificates or opinions. We supply the method and prove it was followed. The judgement is yours.
Both names on the evidence, one name on the opinion
The pre-registration, the harness and the certificate name the body that ran the evaluation and the protocol it was run under. The opinion, the certificate of conformity and the relationship with the provider are the body's alone. A provider or a notifying authority reading the record sees who did the work and what method bound them.
Enforced by the specification, not promised
Antefacts takes no equity in and no contingent fees from any party evaluated under the protocol, does not author criteria for anyone it evaluates, and never scores a competitor. Every commercial relationship between Antefacts and a subject is a published field on the pre-registration, countersigned by the body. Our own group relationships are published on the licensees and disclosures page. The full rules are on the neutrality page.
Mapped to what your assessors already use
All 159 requirements of the protocol are mapped to ISO/IEC 42001, ISO/IEC 23894 and the NIST AI Risk Management Framework on the standards page. Alignment work with ISO/IEC 17065 and 17025 for accreditation readiness is under way.
Fifteen months to choose a method
| Date | Event | What it means for a body |
|---|---|---|
| Aug 2025 | Notified body designation framework live | Applications open; a method must be named in the application |
| 27 Jul 2026 | Regulation (EU) 2026/1744 in force | High-risk dates fixed; the window to choose a method is defined |
| 2 Dec 2027 | Annex III obligations apply | First contestable assessments |
| 2 Aug 2028 | Annex I obligations apply | Product-safety bodies extend into AI |
Bodies choose a method once. Notifying authorities compare bodies to each other, so the method adopted by the first designated bodies becomes the reference for the rest. Our reading of the regulation is in the notes.
How a licence is agreed
A first conversation with your technical director. A read of the specification and the standards mapping by your quality function. A paid pilot: one evaluation in one domain, run by your assessors on the platform, sealed and published as any other. Then a licence, typically nine to twelve months after the first conversation, because that is how long a body's own governance takes and we plan for it rather than around it.
Start the conversation, or ask to read the specification first.