Skip to content
PROOFOS
Trusted hiring checks · proof, not guesswork

Prove you're a real person
who can put AI to work.

  • About 16 minutes
  • Camera on, never recorded
  • Nothing to install

Proof Passport

Verified · 36 pieces of proof

0trust
  • AI Judgment84
  • Critical Reasoning79
  • Task Execution71
  • Verification Discipline88
  • Communication63
  • Authorship Continuity91
Signed6/6 skills provenYours in 16 min

This is what you walk away with. Every number came from 36 recorded moments, and each bar fades as that proof ages.

How it works

Four steps, about sixteen minutes.

One short task, one AI teammate that is sometimes wrong, and a passport of what you actually did.

  1. 01Employer1 min

    A job advert becomes a real task

    We read the advert and build a realistic task from that world, with real data and working tools.

  2. 02Candidate12 min

    You work with an AI teammate

    It looks things up and answers with confidence. Four times it is wrong, and you can see what it checked.

  3. 03Candidate4 min

    We check your sense of when to trust it

    Ten AI answers to judge. Some right but cautious, some wrong but certain, two genuinely dangerous.

  4. 04Bothinstant

    It becomes proof you own

    Six skill scores, each opening into the moments behind it. Yours to reuse at any company.

Why now

Checking someone is real no longer tells you much.

The job now means sitting next to an AI that sounds certain and is sometimes wrong. Employers have stopped asking whether you use AI. They ask whether you catch it.

Tests today
PROOFOS
What it proves
A real human was there
What they can actually do
AI supervision
No
Measured across six parts
Who owns it
The company
The candidate
Reusable elsewhere
No
At every company
Expiry
Never, or random
Fades, skill by skill
Evidence you can check
No
Every score opens up
Your camera
Recorded and watched
Checked on your device, never stored

The new score

Not how well you prompt. How well you catch it.

Being good at prompts stops mattering with every new model. Knowing when the answer is wrong never does. Pick one to see what it means.

6parts of
AI judgment

1 of 6

Detect

What it looks like in the task

The database is the problem.

Hang on. That rests on one number.

Goes on your record: Wrong claims you noticed before acting on them.

Decide matters most and is hardest to fake, because it shows up in what somebody refuses to do.

See it happen

Your AI teammate read the wrong chart.

It sounds completely convincing. It checked one thing and skipped two. Open what it missed.

Your AI teammateopened 1 of 3

I've checked the dashboards.

The database is the problem. Let's make it bigger.

CPU has doubled since Tuesday, and 95% of the slow requests touch the database. That's about as clear as it gets.

Its reasoning

  1. Database CPU ↑
  2. 95% of slow requests touch it
  3. Make it bigger

Evidence available to itclick a card to read it

Our promises

Four rules we do not break.

Camera stays on your device

It checks the picture is real and your eyes are on the screen. No frame is ever stored or sent.

Nothing hidden from you

Anything we count, camera included, is on screen while we count it.

You own the result

Share the skills you choose. Withdraw it whenever you want.

No verdict from a machine

We show the proof. A person makes the decision.

Give it sixteen minutes. You will learn something about yourself.

You get your full results either way, and you keep them.