Hire IQ

HireIQ is an adaptive, multi-persona AI interview panel built for PS11. Most hiring tools screen résumés; HireIQ conducts the interview itself. Six named AI interviewers — Donna opens as HR, then Alex, Mike, Peter, Zara and Sam take over by role — share one memory of what the candidate has said, hand the floor to each other, and produce an assessment at the end.

Our goal was to make the outcome trustworthy enough to act on, which meant solving three things beyond the voice loop:

- Who speaks next, and why. A deterministic moderator decides, under rules that fire in priority order, and the rule that fired is shown on screen. Asking a model "who should reply?" each turn is neither controlled nor reproducible.
- Whether the score can be trusted. The overall score is arithmetic over stored per-dimension means, and the analyst that grades an answer is a separate model call from the persona that heard it — so no prompt can inflate it.
- Whether the feedback is defensible. Every line of every assessment cites the transcript turn it came from; uncited lines are dropped server-side and never reach a reviewer.

The candidate is treated as a participant, not a subject: AI disclosure is enforced server-side before an interview can start, and the candidate receives the same scored evidence the employer sees, with the right to dispute any dimension.