How PsyMetrics AI Test Architect builds a hiring assessment

PsyMetrics AI Test Architect helps employers assemble pre-employment assessments from a job description. Its AI assistant, Psy, identifies role requirements and recommends assessment scales from the company’s existing library, according to the PsyMetrics AI product page. It is an assessment-design workflow—not, by itself, a hiring decision.

The important distinction: quickly building a test is not the same as demonstrating that its scores support a particular hiring decision. HR still needs evidence connecting the chosen measures and scoring rules to the actual job.

Product information checked October 4, 2026. The legal baseline below concerns U.S. federal employment testing rules.

How Test Architect works

According to the PsyMetrics AI product page, Psy extracts competencies, skills and behavioral characteristics from job descriptions, aligns them with the PsyMetrics test library, and selects scales or pre-built assessments. A scale measures a particular attribute; an assessment battery combines several measures into one evaluation.

The company’s September 21, 2026 launch announcement describes the input, recommendations and human review:

1. Supply the role information. Upload a job description or enter role parameters.
2. Receive recommendations. Psy matches requirements against a library the company says contains more than 200 validated scales, covering cognitive abilities, personality traits, workplace behaviors and job-specific skills.
3. Review the composition. Talent leaders review and authorize the proposed combination; the announcement describes this as human-in-the-loop decision support.
4. Deploy the assessment. The custom battery is added to the customer’s account for delivery to candidates, according to the product page, with branding options described in the release.

The company says construction takes minutes and resulting assessments take approximately 20–30 minutes to complete. Those are different clocks: employer setup time and candidate testing time. The completion-time figure is a vendor claim, not a guarantee for every configuration.

For example, an HR team designing an assessment for an inventory role might consider attention to detail, following instructions and relevant math skills. Those appear in the company’s public skills list. This is an illustrative selection—not a verified Psy recommendation or evidence that all three measures belong in every inventory-role assessment.

Test Architect is separate from score analysis

PsyMetrics lists Test Architect and Score Report Analysis separately on its AI product page. The documented Architect workflow concerns choosing and assembling tests. Do not assume its description also explains how a candidate’s answers are scored, how AI interprets results, or whether an employer uses automatic rejection rules.

The wider platform does support selection-oriented features. Its science page describes “Star Profile” benchmarks that flag candidates whose scores fall within an employer’s ideal ranges, alongside score reports containing expected behaviors and interview guidance.

A buyer therefore needs to examine the whole workflow: who chooses the measures, who sets the benchmarks, and what happens when a candidate falls outside them? Human approval of the test composition does not answer those downstream questions.

What the validation and turnover claims establish

The launch release calls the library’s scales “independently validated” and reports employee turnover reductions of 25%–87% across client deployments. These are company claims distributed through a press release—not independent findings from Markets Insider.

The release does not provide enough study detail to evaluate those figures: sample sizes, comparison groups, follow-up periods or the contribution of other hiring and workplace changes. It also does not establish that AI Test Architect itself caused the reported reductions.

For U.S. employers, the EEOC’s employment-testing guidance says tests should be validated for the positions and purposes for which they are used. Vendor documentation can help, but employers remain responsible for ensuring their tests are valid under the Uniform Guidelines on Employee Selection Procedures. “Validated library” is therefore not a complete answer about a custom battery, benchmark or cutoff.

Before HR publishes a custom battery

Use the generated assessment as a proposal requiring review:

• Check the input. Have someone who knows the work confirm that the job description reflects essential tasks, rather than generic preferences such as “high energy.”
• Request relevant evidence. Ask for technical documentation supporting the selected scales, their combination and intended scoring use for comparable work and applicants.
• Document decision rules. Record benchmarks, cutoffs, score weights, reviewer authority and whether low scores trigger rejection or further review.
• Test accessibility and monitor outcomes. Establish an accommodation route, check administration barriers, and examine selection outcomes for adverse impact after launch. If a selection procedure disproportionately screens out a protected group, assess whether an equally effective, less discriminatory alternative is available.

These checks reflect the EEOC’s guidance on job relevance, appropriate scoring, less discriminatory alternatives and disability accommodations. HRaizon’s AI hiring vendor validation checklist provides a broader review framework. Separate AI-specific obligations depend on jurisdiction and the deployed workflow; this is informational, not legal advice.

What candidates should ask

A PsyMetrics invitation does not tell you which scales the employer selected or how much influence the result has. Before starting, ask:

• How long is this particular assessment, and is it timed?
• Which job skills or attributes does it measure?
• How do the results affect progression, and who reviews them?
• Who handles accommodation requests or technical problems?

Under the U.S. ADA, covered employers generally must provide reasonable testing accommodations for the known disability-related limitations of an otherwise qualified applicant unless doing so would impose undue hardship, as explained by the EEOC. Request needed adjustments before testing where possible.

The useful question is not simply whether Psy helped build the assessment. It is what this employer chose to measure—and how it will use your result.