Assessment Methodology

Last updated: 4 August 2026

Dulty measures performance in controlled AI role-play scenarios. It supports, but does not replace, human hiring judgment.

What is measured

A versioned scorecard evaluates observable behaviors such as discovery, rapport, objection handling, commercial judgment, boundaries, and closing. Evidence links to exact turns in the completed simulation.

How to interpret results

Scores and AI verdicts are estimates from a limited simulation. Confidence, strengths, risks, and evidence should be reviewed together. Managers record a separate decision and remain accountable for it.

Fair use

Use the same template, language, and difficulty when comparing candidates. Offer reasonable alternatives where accessibility or language would otherwise disadvantage a candidate. Do not infer protected characteristics.

Limitations

Model output can be incomplete or wrong, and performance may differ from real work. Reassess after material methodology changes; scorecard versions are retained for traceability.