Unproctored delivery is disclosed
The AI Adaptation Profile and the Equilibrium instruments are delivered without a human proctor. That fact is disclosed to participants before they begin, and the results are interpreted with that condition in mind.
Conformance
Standards and conformance
How the Equilibrium service is operated, what standards we align ourselves with, and what we do not yet hold. We say the quiet part out loud because procurement teams and participants both deserve the same facts.
ISO 10667
Assessment service delivery
The service is operated in conformance with ISO 10667-1 and ISO 10667-2, the international standards for assessment service delivery. In plain English, that means we have documented how the whole service is planned, delivered, reviewed, and improved, and we hold ourselves to those commitments. The standards cover the following areas.
The agreement with the client
Every engagement starts with a written agreement that records the purpose of the assessment, the population it will be used with, the instrument version and norm version, the roles and responsibilities of each party, and the cadence at which the service will be reviewed. We do not begin processing until the agreement and the corresponding participant-facing notice are aligned.
The rights and obligations of the assessment participant
Participants are told the purpose of the assessment, who will see the results, and how long data will be retained, before they begin. Participation is voluntary, and consent is purpose-specific and versioned. Participants can ask for access to their data or request deletion at any time; deletion is carried out by crypto-shredding so the audit record remains intact while the person's data becomes unreadable.
How assessment data is handled and retained
Assessment data is stored in-region, encrypted in transit and at rest. Retention periods are agreed in writing and can be shortened on request. Individual item responses are never visible to an employer or manager; only aggregates are reported, and only when at least five valid responses exist for the group. Response-quality checks are run before any score is produced, and sessions flagged as suspect are excluded from norms and team analytics.
How results are delivered and interpreted
Results are delivered as an advisory report, never as a decision. Each reported score is drawn with a likely range derived from the standard error of measurement, and the platform does not describe two people as different when their bands overlap. Reports include a clear statement that the results are not the sole basis of any decision about a person, and must be combined with other information and human judgement.
How the service is reviewed
The service is reviewed periodically against the agreed purpose, the latest version of the instrument, and the quality of the data collected. Item performance, response quality, and fairness indicators are monitored. When an update to the instrument or a norm is released, the version used for each report remains traceable, and any comparison against a prior version is flagged and explained.
This is a declaration of conformance made by us. It has not been certified by a third party, and we say so because the distinction matters.
ITC
Technology-based assessment guidelines
We also align ourselves with the International Test Commission guidelines for technology-based assessment. Our specific commitments are:
Results are never the sole basis of a decision
No individual report or team aggregate is designed, configured, or described as the sole basis of a hiring, promotion, termination, or other decision about a person. The platform enforces this at the point of interpretation.
Accessibility is considered
The interface is built to WCAG 2.2 AA, works on mobile and desktop, and is keyboard-navigable. Plain-language alternatives are provided for technical terms. We do not introduce time pressure for the standard item blocks, and we continue to test with assistive technologies.
Technology does not introduce construct-irrelevant variance
The device or browser a person uses does not affect the score. Response timing is used only for quality checks, never for scoring. Language models are used for narrative generation only after the scores are fixed, and every generated statement is checked against the underlying numbers before it is shown.
What we do not yet hold
Honest gaps
Under-claiming is the brand. We will not borrow credibility we have not earned. As of today, the following are not held by Equilibrium or AdoptAI.
No third-party ISO 27001 certification
We operate security controls consistent with ISO 27001, but we have not yet been certified by an independent third-party auditor. We will publish that certification when it is true and not before.
No ISO 42001 certification
Our AI governance practices are documented and reviewed, but we do not yet hold ISO 42001 certification for an AI management system. We will publish it when it is true and not before.
No published criterion validity study
We have not yet published evidence that these scores predict job performance, retention, career change, or any other real-world outcome. Criterion-validity work is in progress with design partners, but it is not complete. We will publish it when it is true and not before.
No British Psychological Society test registration
Equilibrium is not registered with the British Psychological Society's Psychological Testing Centre. We will register it if and when it meets the requirements and we choose to seek that status. We will not claim it before then.
Data handling
What happens to assessment data
Assessment data is stored in Australia and processed in-region. We use encryption in transit and at rest, and access is logged. Retention periods are set out in the client agreement and can be shortened on request. Individual item responses are never visible to an employer; team views only render when at least five valid responses have been collected for that group. Reports that would fall below that threshold are suppressed, not rounded down. The full privacy policy is in our privacy policy.
- Stored in-region and encrypted at rest and in transit.
- Retention is configurable per client agreement and region.
- Five-response suppression floor for any team analytic.
- Individual item responses are never visible to an employer.