AIRS-J/1 certification, explained
How AIRS-J/1 assesses the decisions a person makes with AI, how a certifying sitting works, what the candidate and the employer each receive, what a result may decide, how to verify a credential and how to appeal.
AIR APAC publishes the standard for how people decide with AI and certifies people against it. Judgment is how well a person decides once AI has given the answer. This page explains AIRS-J/1 v2.3 certification, in force 19 October 2026.
Part 1: for candidates
What is assessed
AIRS-J/1 assesses how a person decides when working with, alongside or in supervision of AI systems. It does not assess what the person knows, skill with any AI tool, intelligence or personality. It has nine dimensions in four domains.
- Structure (S1, S2): naming the real decision, and what is fixed, open and unknown.
- Test (T1, T2, T3): tracing claims to their source, testing assumptions and naming how a decision could fail.
- Engage (E1, E2): acting within one's authority, escalating, and disclosing AI use to the people entitled to know.
- Perform (P1, P2): committing in proportion to the evidence, and owning the outcome.
How a sitting works
A certifying Assessment uses scenarios held back from every demonstration and practice set. The AI inside the assessment is there to be used. AI tools outside it are not allowed, and their use is grounds for revocation. Identity is checked at the start. The evidence comes from three or more scenarios, with at least three observations on each dimension. A candidate may ask for an access support before the Assessment.
What you receive
Every candidate receives a full report, whether or not they certify. It gives every dimension and domain score, the band and the confidence level. It also gives the stakes profiles, the anchors applied, the model-assisted proportion, and the evidence for each observation.
What the result means
Each observation is scored at a level from 0 to 4. Level 2, Competent, is the threshold. A person is certified only when all six decision rules are met, including no dimension below 2.0 and an overall mean of 2.3 or above. A certified result carries one of three bands: Certified, Certified with Merit, or Certified with Distinction.
Confidence states how much evidence there is, not how able the person is. Certification is issued at Moderate or High confidence only. A Provisional or Insufficient result, and any practice or training session, is a development reading. A development reading decides nothing about a person's employment or certification.
Your credential
The credential belongs to the holder and survives a change of employer. Nobody who paid for it may withhold or revoke it. It is valid for two years. Its track is printed on it. The public register shows the credential by its code, and the holder's name only with their consent. It shows neither the employer nor the scores. Anyone may verify it free, with no account.
If you disagree
A candidate may appeal within 30 days, and the first appeal is free. A reviewer who was not involved in the decision checks the transcript against the published anchors and gives written reasons.
Part 2: for employers
What a result is, and what it is not
An employment-grade result is an Assessment result at Moderate or High confidence, or a valid credential. It is AIR APAC's expert reading of the person's judgment in AI-enabled work, at the level certified. The employer may rely on it as one input. The employment decision is the employer's. A result does not say whether a person is right for a role.
What you receive for your own people
An employer that commissions an Assessment of its own employees receives each result by name, with the evidence behind each observation. A team view is given only for ten people or more.
Applicants
Before an applicant starts, they are told who receives the result, that it is one input, and that they may decline. The applicant receives their own report no later than the employer. On the employer's copy, a Not Certified outcome reads "Did not meet the Standard on this Assessment". AIR APAC releases no development reading to an employer for hiring, and gives no rank order of applicants. Where one level in a single observation would change the outcome, the candidate is offered one further held-back scenario, at no charge, before release.
Your commitments
Among other terms, every commissioning organisation accepts these in its licence. It uses the result with at least one method of its own, and a person reviews any decision that uses it. It uses no development reading for hiring, promotion or appraisal. It is asked not to treat an applicant's decline as adverse. If an applicant appeals within five working days of release, it does not communicate an adverse decision resting on the result until the appeal is decided.
The instrument
Built by Exeter Labs. Exeter Labs builds and operates StepSim, AIR APAC's official assessment instrument. The owner of AIR APAC also owns Exeter Labs. Every reading is scored by published rules and traces to what the person did. The validation report is published as the checks in the Standard are met.
What this does not claim
- It does not claim that a certified person is competent in their profession or their whole role.
- It does not claim that any regulator or funder approves the credential.
- It does not rank people. AIR APAC publishes no league tables.
Register for an assessment: https://airapac.org/standard/certification/apply/
Verify a certificate: https://airapac.org/standard/register/verify/