Every score cited · CSV in, ATS out.
AI used in recruitment is named explicitly as high-riskunder the EU AI Act's Annex III — impact assessments, human oversight, candidate notice, and logging, all binding from 2 August 2026. We built for that bar from day one: every score cited to a timestamp and a quote, integrity flags kept private with disclosed false-positive rates, and consent before the microphone ever opens.
·A real reply from the founder, not a sales team ·Candidate consent & data handling built in · Bias-resistant by design
Recruitment AI is named explicitly as high-risk. From 2 August 2026: fundamental-rights impact assessments, human oversight, candidate notice, technical documentation, logging, and post-market monitoring — or penalties up to €35M or 7% of global turnover.
Independent bias audits, published publicly, at least once a year — and it applies extraterritorially to any employer evaluating a candidate who resides in NYC, wherever the role or the company sits.
One more state with its own hiring-AI disclosure and impact-assessment regime already in force — the pattern isn't slowing down, it's compounding.
Generic AI screeners run a 30–50% word-error rate on Indian-accented English versus 2–8% on native speakers — a measurable, auditable disparate-impact vector, not a rounding error. Under an EU AI Act impact assessment or a Local Law 144 bias audit, that gap is exactly the kind of finding that ends a vendor relationship.
Your senior engineers spend it on first-round phone screens that a structured interviewer could run identically — and they're inconsistent every time anyway.
Across a 19,368-interview dataset. Copilots feed candidates answers your panel can't see. You need integrity that's verifiable, not vibes.
Generic AI screeners mis-transcribe Indian-accented English and quietly penalise it. That's not a fit signal — that's bias you can be audited for.
Drop a CSV of candidates and the JD — or your own question bank. We build the role blueprint: 15–20 weighted, time-budgeted topics.
Each candidate gets the identical structured voice interview, on their own time. No scheduling, no no-shows, one bar for all.
Every score is cited to a timestamp, a quote, a code run, a diagram element. Integrity checks run quietly in the background.
200 candidates in, 200 fit reports out — pushed straight to Greenhouse, Lever, or any ATS. You open a ranked shortlist.
Structured assessment mode runs the same blueprint, the same topics, the same bar for everyone — so you're comparing candidates, not interviewer moods.
Optional presence and eye-tracking checks for high-stakes screens. Integrity signals are kept private, surfaced as a flag for your team — never as a public verdict on the candidate.
No “the AI thinks they're a 7.” Each rating links to the exact timestamp, quote, code run, or diagram it came from — agreeing with your own interviewers ≥85% of the time.
The accent handling that protects candidates is also what protects you: scores reflect what was said, not how it was accented — reducing a bias other tools bake in.
Drop 200 candidates, get 200 fit reports mapped to your role-asks, pushed into your ATS. Over time, every candidate who's demonstrated a role becomes part of a verified pool you can hire from directly — screening you only have to do once.
Your numbers, not ours. We only assume the ~45 min/screen your own seniors spend today.
Same bar for every candidate, evidence behind every score, and your engineers out of the phone-screen treadmill.
Run a consistent first round at volume, free your engineers from phone screens, and hand hiring managers a ranked, evidence-cited shortlist.
Screen a large bench fast and submit candidates with a defensible fit report attached — proof of quality your clients can trust, not just a CV.
Give every student a real, structured mock under one bar, show placement partners verified readiness, and route the strongest into the talent pool.
Proctoring done badly punishes nervous, neurodivergent, or differently-accented candidates. We treat integrity as a private signal for your team, with the limits stated up front — not a verdict stamped on a person.
Same honest mechanic — one credit, one full structured interview and fit report — bought as a monthly pool your team draws from. No per-seat tax, no shelfware.
Illustrative pricing · set your own. Pilots are scoped to one role so you can judge the fit reports on real candidates.
Run one role. Judge the fit reports.
A pilot is scoped to a single role on real candidates — so the decision is the work, not the pitch. CSV in, shortlist out.
· Scoped to one role ·Consent & compliance built in · A real reply from the founder