Each skill measures an axis the models struggle to reproduce. Candidate scores are pitted against the latest versions of the public models, on the same challenges.
A résumé won’t tell you whether someone can infer a rule from three examples, stay sharp when the environment changes its rules, or read the real emotion behind a polite message. These are exactly the grounds where today’s AI models stumble — and therefore where a human still makes the difference. Our challenges measure these skills directly, through real scenarios, not a self-reported questionnaire.
6 skills available · 6 challenges in the catalog
A logic test where you infer a rule from just a few examples. Measure your ability to abstract complex rules.
Faced with an emotional dilemma, pick the most fitting response — where there is no obvious right answer.
Identify the emotion truly felt and its real cause, beyond the surface signal.
How confident are you in your own answers?
The candidate negotiates 5 successive purchases against an LLM seller, aiming to capture as much margin as possible while staying under budget.
Each skill is a concrete exercise: infer a hidden rule, negotiate five purchases against a seller model, pick the right reaction in a dilemma with no obvious answer. We observe what the candidate does, not what they claim they can do.
Every challenge is calibrated against a panel of real profiles. A raw score means little on its own; set against that panel it becomes readable — “better than X% of candidates on this skill” rather than an abstract grade.
The same challenges are run by the latest public models. The candidate sees where they stand, and above all where humans still hold the edge over the machine — which is the whole point of hiring for these skills.
Automation absorbs the repetitive, codifiable tasks. What gains value are the skills models reproduce poorly: abstraction from few examples, adaptation when the rules shift, fine emotional judgement, calibrated intuition under uncertainty. Assessing these axes means hiring for what will stay distinctly human tomorrow.