Adoption Levels
AIEP adoption is not binary. There is no minimum threshold before you can publish an artefact, and no ceiling above which you cannot go further. The protocol defines three meaningful adoption levels, each building on the last.
The three levels
| Level | Name | What you publish | What you gain | Typical effort |
|---|---|---|---|---|
| 1 | Discoverable | index.json and metadata.json at /.well-known/aiep/ | AI retrieval systems can find you, identify you, and enumerate what you have published | Less than a day |
| 2 | Verifiable | Evidence-referenced artefacts matching canonical schemas, with hash fields populated | Retrievers can validate provenance — who published it, when, and against what evidence | A few days, depending on artefact count |
| 3 | Certified | Registry listing with DID, certificate artefacts, compliance signals, and audit log availability | Machine-verifiable “AIEP Certified” claim; conjunction trust model becomes active | Weeks for full certification cycle |
Level 1 — Discoverable
Publishing the two required discovery files at /.well-known/aiep/index.json and /.well-known/aiep/metadata.json makes your domain discoverable by any AI system configured to walk AIEP Mirror endpoints. No registration, no application, no fee.
At this level you are telling retrieval systems: I have published structured knowledge here. They can enumerate your artefacts, but cannot yet verify individual claims.
See Quickstart for the minimal file structure.
Level 2 — Verifiable
At Level 2 you begin publishing artefacts that conform to at least one of the 13 canonical AIEP schemas. Each artefact carries:
issuer_did— who published itevidence_refs— what supports the claimartefact_hash— the content hash at publication timeschema_version— which schema revision this conforms to
A schema-conformant artefact with a populated artefact_hash can be independently verified by any retriever. This is the point at which AIEP adds more than discoverability — it adds provenance.
See the Schema Catalogue for the full list of schemas.
Level 3 — Certified
Level 3 requires a registered issuer DID in the AIEP Registry and the publication of one or more certificate artefacts. At this level:
- Your identity is anchored to a DID Document resolvable via the registry
- You can make the “AIEP Certified” claim in your own artefacts
- Third-party verifiers can check your certification status without contacting you
- Conjunction trust becomes possible: your artefacts can participate in multi-party evidence chains
Certification is not a one-off event. It requires periodic resubmission and carries an expiry. Fail-closed behaviour applies: an expired or revoked certificate disqualifies the claim, not downgrades it.
See Certification for the full certification process, and Access Tiers for the governance capabilities that unlock at each tier.
What level should I target?
| If you are… | Start with level… | Why |
|---|---|---|
| A solo developer publishing a dataset | Level 1 | Discoverability costs nothing and gets you into AI retrieval pipelines immediately |
| An organisation publishing research outputs | Level 2 | Verifiable artefacts are the minimum for third-party trust |
| A regulated entity or enterprise compliance function | Level 3 | Conjunction trust and machine-verifiable certification are non-negotiable at this scale |
| Building an AI system that retrieves from the web | Level 2 (as a consumer) | You should check artefact_hash and evidence_refs before trusting a claim |