The Screening Agent writes a scoring blueprint from your job description — no canned rubric — then scores every applicant across technical, core, behavioural and evidence pillars and sorts them into advance, review and reject at thresholds you set per job. It calibrates those thresholds against who you actually hired, and reports disparate impact both for its own raw calls and after your team’s overrides, so you can prove the bar was fair.
It reads the whole job description and generates the scoring dimensions and their weights for that specific role, rather than forcing every job through one fixed template. Résumé, interview and assessment signals then roll into a single composite with a verdict and a confidence level.
It compares the scores of people you actually hired against people you rejected and proposes new thresholds when the gap is real. You approve the change — it never silently moves the bar.
It reports advance, review and reject rates by group with a disparate-impact ratio, shown both for the agent’s raw calls and after your team’s overrides, and exports every decision as a CSV for audit.
And writes the scoring dimensions and weights.
Same bar, same evidence format, every candidate.
Advance, review or reject — and why, in writing.