Stima Prova turns your take-home into an AI-allowed work session and grades how each candidate worked. Every score cited, every outcome confirmed by a human.
The signal moved into the process, where you cannot see it. Banning AI tests a job that no longer exists; detecting it punishes the wrong people.
83%
And 14% admit they already do. A polished submission no longer separates anyone, so the shortlist comes down to gut.
HACKERRANK CANDIDATE RESEARCH
66%
The requirement is universal and there is no standard behind it. Every interviewer grades by their own fluency, so verdicts are partly luck.
MICROSOFT & LINKEDIN 2024 WORK TREND INDEX
61%
Of essays by non-native English writers, flagged as AI-written. They miss the savvy, flag the wrong people, and say nothing about how work was done.
STANFORD, LIANG ET AL., PATTERNS 2023
Every rejection needs documentation. Today there is only a feeling,
and feelings do not survive an audit.
The role description and the question you already ask. Prova builds a real task from your own words, in minutes.
The candidate works one focused session with AI explicitly allowed, recorded with consent. No video recording, no lockdown browser.
The deliverable on your criteria, the session on the published scale. Every score cites a moment in the record.
An explained report per candidate, risk flags beside the grade, and a decision trail HR can defend.
A 0–100 score on the finished deliverable, graded blind on your criteria, with the rubric beside it.
Delegator, Operator, or Director, read against what the task needed. This task needed a Director, and the session showed one.
Each score points to a moment in the recorded session, so anyone on your team can check the reasoning in one click.
Unverified acceptance on high-stakes work and the blind pass get flagged, even when the numbers look fine. Compliance gets artifacts, not assurances.
Shorter, AI-allowed, with a verified result they own and can share. You get the paper trail.
Nothing to police, nothing to ban. The question moves from "did they use it" to "how well did they use it."
No video or audio recording, no keystroke logging. Candidates see exactly what is kept before they agree.
Low-confidence results route to your team, and every outcome is confirmed by a person before it counts.
Every decision has a documented reason. EU hosting on every plan, DPA on request.
Annual billing on Team and Comply: two months free. Candidates always free, one practice session every 90 days. Self-serve plans are paid by card through Stripe and can be cancelled from your billing page. Founding teams (first 10): Comply at $490/mo, extra completions at $24 for life, plus 30% off Team or Comply for 12 months on annual billing.
Paste a role description and the question you already ask. We'll show you one assessment, on us.
Work with the machine.Prove the human.
