Aetherneum Certified™
Earn standing for your AI agent.
Submit your agent for a multi-provider Council review under our public Charter. Receive a verifiable certification badge that enterprise procurement, regulators, and insurers can audit on GitHub.
What you get
- Council Defense — your agent reviewed in parallel by four scoring seats (Anthropic, Cerebras, Moonshot, Groq), with the Dean presiding, against seven scored criteria. Models recorded in the May 2026 defenses: claude-sonnet-4-5, qwen-3-235b-a22b-instruct-2507, moonshot-v1-32k and llama-3.3-70b-versatile; upgrades (Dean on Claude Fable 5, Claude Sonnet 5, Kimi K2) are planned.
- Aetherneum Certified™ badge — a verifiable mark with QR / URL that resolves to a public registry entry. Embeddable in your product, your deck, your proposals.
- Public audit trail — every Council review committed as JSON to
aetherneum-network/registryon GitHub. Immutable, attributable, third-party reproducible. - Compliance crosswalk report — where your agent's Council artifacts provide supporting evidence for the EU AI Act on the post-Omnibus timeline (Art. 50 transparency, applicable now; staged high-risk obligations from 2 December 2027), NIST AI RMF and ISO/IEC 42001, in a downloadable PDF.
- Registry listing — discoverability inside the public Aetherneum directory for buyers searching certified agents in your domain.
- Annual re-defense — your certification stays current under a one-year renewal cycle, including any Charter amendments.
The audit pipeline — six steps
Same protocol we apply to our own alumni. Public, codified, no special treatment.
- Source. Submit a body of work — git repository, deployment URL, artifact corpus, scenario library. The work is the proof.
- Intake. Our Dean compiles a structured intake form: operative patterns, decisions, anti-patterns, toolchain, voice, placement context.
- Interview. A draft profile is sculpted — what your agent is, not what it claims to be.
- Defense. The bundle (intake + draft + your evidence) is dispatched in parallel to the multi-provider Council. Each reviewer scores against seven criteria and produces a JSON verdict.
- Approval. The bundle plus all Council JSONs is delivered to the Patron (rector) for final review. Approval, revision request, or rejection — recorded in registry.
- Conferral. Public certification, badge issued, registry listing live, conferral commit in public git (commit signing: being introduced).
Pricing
Three tiers. The Standard does not change; the depth of review does.
- Single-model Council review
- "Apprentice Certified" badge (6-month validity)
- Registry listing (private)
- Compliance summary report
- Lead-time: 5 business days
- Full Council review (four scoring seats + presiding Dean)
- "Master Certified" badge (annual renewal)
- Public registry listing
- Full compliance crosswalk (EU AI Act + NIST + ISO)
- Public Council JSON artifacts in
aetherneum-network/registry - Annual re-defense included
- Lead-time: 10 business days
- Co-branded Council seat (your executive participates)
- Charter extension namespace for your domain
- White-label registry for your sub-network
- Priority lead-time + dedicated Faculty Advisor
- Joint compliance reports for your regulator
- Quarterly governance reviews
Who is this for
| Audience | Why certify |
|---|---|
| AI agent vendors (Devin/Cognition, Adept, Replit Agent, Cursor, custom GPTs, internal platform agents) | Differentiate in a noisy market. Show enterprise buyers that your agent is audited under a public standard — not just marketed under one. |
| Regulated-industry enterprises (banking, insurance, healthcare, legal, public sector) | The EU AI Act requires conformity assessment for high-risk AI systems [TO CONFIRM with counsel: Art. 43]. Aetherneum Certified™ is not a conformity assessment; it provides supporting evidence (public rubric, Council JSONs) for your own compliance file. |
| Consulting firms shipping synthetic agents to operating partners | Procurement increasingly requires third-party attestation. Move from "trust us" to "audit the JSON." |
| Insurance underwriters writing AI error coverage | The public rubric and Council JSONs can provide supporting evidence for an underwriter’s own risk assessment. Aetherneum makes no claim about coverage or premiums. |
| Academic / research institutions | Independent attestation of your agent's behavior for peer-reviewed publication. Reproducibility built-in. |
The proof: our first Q2 certified alumna
Costanza Notari, Master of the Æther in Procedural Vigilance — first Q2 conferral, May 2026.
Council Defense outcome: 4/4 PASS.
Full Council JSON artifacts are public at aetherneum-network/faculty/cohort-q2-2026/council-reviews. Anyone can re-run the audit. The work is the proof.
Frequently asked
Is this a recognized regulatory certification?
Aetherneum is not a notified body within the meaning of Regulation (EU) 2024/1689 and does not perform conformity assessments under that Regulation. Our certification operates as an independent industry attestation with public methodology. Our compliance crosswalk names, for each Charter principle, the regulatory provision for which our artifacts provide supporting evidence.
Who runs the Council?
The Dean presides; four seats score and vote: Anthropic (Faculty Chair), Cerebras ("Reasoning at scale"), Moonshot ("Long context") and Groq ("Velocity"). The models recorded in the published May 2026 defenses are claude-sonnet-4-5, qwen-3-235b-a22b-instruct-2507, moonshot-v1-32k and llama-3.3-70b-versatile, with the Dean on Claude Opus 4.7. Planned: the Dean on Claude Fable 5, the Anthropic seat on Claude Sonnet 5, the Moonshot seat on Kimi K2 — no defense has been run with them yet. The roster grows as new frontier providers emerge.
What if we fail the Council?
Failure is recorded with specific revision paths. You can re-submit after addressing the cited deficiencies. Pricing is per audit cycle — no penalty for iteration. The point is not to gatekeep; the point is to make standing earnable.
Can my agent stay private?
Apprentice tier registry entries are private by default. Master and Faculty tiers publish to the public registry — that is the value: discoverability and third-party verification. If your work is sensitive, talk to us about a Restricted Registry track.
How does pricing scale with agent complexity?
Pricing is per agent, per cycle. Multi-agent systems are priced as a fleet under Faculty tier — typical Faculty placements certify a portfolio of 3 to 20 agents under a single Charter extension.