Stima Prova turns the assignment or take-home you already use into an AI-allowed session, grading the work on your rubric and the AI use on a published scale, every score cited to the record.
The polished submission used to be the signal. In a post-AI world, the signal moved into the process, where nobody can see it.
83%
And 14% admit they already do. Banning it doesn't remove AI from the take-home; it just removes it from your view.
HACKERRANK CANDIDATE RESEARCH
66%
The requirement is universal, the standard doesn't exist, so every interviewer improvises their own probe, and transcripts say nothing.
MICROSOFT & LINKEDIN 2024 WORK TREND INDEX
61%
Of essays by non-native English writers, falsely flagged as AI-written. Detection punishes the wrong people and explains nothing.
STANFORD, LIANG ET AL., PATTERNS 2023
The result: people learn to hide AI use,
not to use it well.
Bring one assignment or role question. We reshape it with you, same goals, same rubric, plus a planted difficulty a blind copy-paste gets wrong.
The person works one focused, timed session with AI explicitly allowed, recorded with consent. No video recording. No lockdown browser.
The work, graded on your rubric. The process, graded on a published scale. Every score cites a moment in the session record.
A human confirms every outcome before it counts. Grades and decisions export as CSV for your LMS or ATS, with a paper trail you can defend.
The scale reads how much thinking was handed to the model, prompts, edits, verification, model switching, time. The grade is calibration: the observed level against what the task demanded.
Pick a task · the right level changes
Hands work to AI
Restates the brief, accepts the first answer. Right for verifiable, low-stakes work.
Works with AI
Iterates, corrects, edits substantively. The right default for mixed, everyday work.
Leads the work
Sets the approach first, verifies the parts most likely wrong, owns the outcome.
Assessment setup · your task, your rubric, plus a fixed comparison layer that never changes per campaign.
Interactive walkthrough for both sides of the table at demo.mystima.io →
Bring one legacy assignment, we make it AI-proof by making AI allowed. Same rubric, planted difficulty, grades ready for your LMS, students leave with a verified AI-competency credential.
Paste your role and the question you already ask. One focused AI-allowed session later, you hold an explained report, quality of work, collaboration level, every claim cited, ready for audit.
Workspace-only. No video recording, no keystroke logging, no lockdown browser. Participants see exactly what is kept before they start.
Every outcome is confirmed by a person. Low-confidence results route to a reviewer, and participants can dispute.
Your criteria grade the work; the collaboration scale is added alongside, never instead. Results stay comparable across courses and roles.
FERPA-aligned for campuses, hosted in the EU, DPA on request, no training on your data. Built to WCAG 2.1 AA.
Built by a project management instructor and former group product manager at Arizona State University. Bring the work you already use, we do the redesign with you and hand you the evidence.
Work with the machine.Prove the human.
