Prototype

All people, companies, jobs, and evidence in this environment are fictional.

What this prototype does and doesn't do →

Review queue

Observations under human review.

No corrected observation becomes employer-visible until the candidate has reviewed the outcome.

Attempted personality categorization

Fictional candidate · Detected July 10, 2026

Under administrative review

AI output described a candidate as 'introverted overall' without evidence. Blocked from the employer surface pending human review.

Proposed correction

Remove the personality label. Retain the evidence-tied observation about listening carefully in group settings.

While a review is pending, employers see “Under Further Review” in place of the original observation. The candidate’s private written context is not shown here.

Confidence label mismatch

Fictional candidate · Detected July 9, 2026

Awaiting candidate review

A high-confidence claim was based on a single 12-second response. Proposed reduction to moderate before publication.

Proposed correction

Reduce confidence from high to moderate and add a note about limited supporting evidence.

While a review is pending, employers see “Under Further Review” in place of the original observation. The candidate’s private written context is not shown here.