AI transparency statement

Our matching engine parses a role brief into skills, seniority, market, work mode and hard constraints, then scores candidate profiles against those attributes. The output is a ranked list with a per-candidate rationale showing which attributes drove the score.

It is a ranking tool. It orders a queue that a human then reads.

It does not reject anyone. It does not send or withhold a candidate from a client. It does not score personality, appearance, video interviews, or any inferred protected characteristic.

A named consultant reviews the ranked list — including candidates ranked in the bottom band, on a sample basis — and makes every decision about who is put forward.

We audit shortlist composition quarterly against the applicant pool and track offer acceptance among candidates the model ranked in the bottom half. If that number is healthy, the model is discarding people we would have hired, and it gets re-tuned.

We publish the feature list the engine scores against to any client who asks.

If you have been assessed by our system you may ask what role it played in any decision affecting you, ask for a human review, and ask for the data held about you. Write to [email protected].

Where the EU AI Act applies, these systems are treated as high-risk and we operate them accordingly, including logging, human oversight and candidate notification.