Measure AI judgment
across your team
and your candidates.

Choose the role and send

Select the role to assess, enter candidate emails and set the time limit. Each one receives a unique link.

The candidate designs

Without creating an account, the candidate completes an AI system design form for their role. There are no single correct answers — judgment is what's being evaluated.

You see the results

Global score, per-stage breakdown with feedback, and exactly what the candidate designed. All in your evaluator dashboard.

Global score

A 0–100 index reflecting overall design quality, comparable across candidates in the same role.

Per-stage breakdown

Specific feedback for each of the 6 design stages: objective, capabilities, criteria, restrictions, autonomy, and oversight.

Full design

You see exactly what the candidate designed — not just whether they got it right, but how they thought through the system.

Not multiple choice.
Real judgment.

TestGorilla and similar tools evaluate whether someone knows concepts. Zepai Assess evaluates how they decide when there's no single right answer — exactly what matters when working with AI.