Hiring Decision Panel — by an AI council
Make a clear-eyed hire / no-hire call on a candidate. Four rival AIs debate it and hand back: A hire / no-hire recommendation with the risks and the open questions.
A hire / no-hire recommendation with the risks and the open questions.
Signed off by an independent Final QA audit — with a “verify before you rely on this” list, never a black box.
Why this team
Hiring calls made warm — after a likeable final interview — are where the expensive mistakes live, so this panel is built to cool the decision: the People Lead checks the candidate against what the role must deliver rather than the wish-list, the Data Skeptic separates interview evidence from interview vibes, and the Steelman argues the strongest honest case for the candidate so scepticism does not curdle into rejection-by-default. The Devil’s Advocate names the bias in the room — halo, similarity, seat-filling desperation — and the Pragmatist prices the alternative: what another month of searching actually costs.
Each seat runs on a different frontier model — GPT, Claude, Gemini, Grok — so it’s a genuine cross-model review, not one AI in costumes. How a run works →
What the debate sounds like
An illustrative excerpt — A ten-person startup is deciding on a head-of-sales candidate who interviewed brilliantly, comes from a much larger company, and whose references were warm but curiously vague on quota specifics.
Inventory the actual evidence: one polished narrative, charm in four interviews, and references that praised character while declining to confirm numbers. Vague-on-quota references are not neutral — specifics are easy to give when they are good. The evidence supports “excellent presenter”, which is a sales skill, and says nothing yet about “closes at a startup”. Those are different jobs.
Two biases are steering this room: the halo — interview brilliance colouring every other read — and big-logo transfer, crediting the company’s machine to the individual. At the large firm, leads arrived qualified, and the brand opened doors. At ten people, there is no machine. The panel should also admit the seat has been open four months; desperation is the third interviewer.
The honest case for hiring: senior sellers from large companies bring process discipline startups lack — pipeline hygiene, forecasting, enterprise etiquette — and the reference warmth was genuine even where numbers were withheld. Presenting brilliantly to a hostile panel is the job when the customer is a sceptical CFO. If the missing evidence is closing ability, that is testable — it is not grounds for rejection, it is grounds for a test.
Questions
What should I paste in about the candidate?
The role’s actual deliverables, the interview signal round by round (including doubts you feel disloyal writing down), reference notes, and the state of the alternative — other candidates or the cost of the open seat. The doubts matter most: the panel exists to examine exactly the things the room has been politely not saying.
Is “get more signal” just indecision with extra steps?
Only if the signal is unnamed. The verdict format requires the specific test and the question it answers — a working session, a targeted reference question, a portfolio deep-dive — with a deadline. That converts a stalled maybe into a designed experiment, which is the opposite of indecision.
Can this work for internal promotions and founder hires too?
Yes — the failure modes just change costume: for promotions, loyalty and tenure play the halo role; for founder-adjacent hires, chemistry does. The bias check and the evidence-versus-vibes separation run identically, and the “cost of a wrong hire versus an open seat” calculus is if anything sharper.
Your material is used only to run your review — never to train public models. Encrypted in transit and at rest. Security & privacy →
Want full control — pick your own minds, set the depth? Open the full council →