The honest position on AI in hiring

AI is very good at structure, coverage and speed. It is not good at judgement, and the moment it is allowed to make a hiring decision you have both an ethical problem and, in the EU, a legal one.

Everything built here follows one rule. The machine can advance or flag. Only a person rejects, shortlists or presents.

Modernise recruiting, keep the human touch.

The audit

Four to eight weeks. Every stage of your hiring process mapped against time spent, decision quality and candidate experience. The output is a ranked list of what to automate, what to leave alone, and what to fix before automating anything, because automating a broken process just breaks it faster.

Screening agents

Structured, competency-anchored screening that runs when the candidate is available rather than when your recruiter is. The same questions for every candidate on a dream job opportunity, which is both the fairness argument and the data argument. Scores are weighted per role rather than per template, and every score is reviewable and challengeable.

Stella, the AI screening assistant we use inside our own recruitment process, is this in production on our own dream job opportunities. We do not sell a workflow we have not run ourselves.

Governance

GDPR lawful basis captured at the point of collection, not retro-fitted
Data minimisation by design, with a stated retention period and an automated purge
No automated rejection, which keeps you outside the highest-risk classification under the EU AI Act
An audit log that records what the model saw, what it scored and who reviewed it
Model and prompt versions recorded per screening, so a decision from March can be explained in November

Training for TA teams

An eight-hour instructor-led programme in which your recruiters build five working workflows on your own live roles. Not a demo, not a webinar. People leave with things that run on Monday. The whitepaper sets out the curriculum and the economics.