Vettika
A voice AI runs 12-minute adaptive screens, scores against your rubric, and puts a report in your inbox 60 seconds after the candidate hangs up. Candidates keep their transcript. No black-box rejections, no proctoring overlay.
The problem
First-round screening is the most repetitive hour in hiring and the least fair to candidates. Recruiters lose days to scheduling; applicants get rejected by systems they cannot see. One-way video tools solved the recruiter half and made the candidate half worse.
What I did
- Framed the product around a two-sided promise: recruiters get their time back, candidates get their transcript and score.
- Specified the adaptive interview loop: a real conversation with follow-ups, not a fixed question list read at a camera.
- Defined rubric-based scoring so every candidate is measured against the same stated criteria, and made the report shareable with the candidate by default.
- Built compliance into the spec from day one — NYC Local Law 144, the EU AI Act, GDPR and Nigeria’s NDPR — rather than retrofitting it, and made audio destruction on hangup a product guarantee rather than a policy line.
- Chose pay-as-you-go with three free interviews, no card and no subscription, to remove the trial barrier for the founders, recruiters, talent scouts and hiring managers the product is aimed at.
Where it landed
12 min
average first-round call
60s
from hangup to a scored report
3
free interviews, no card, no subscription
Stack & tools
Next.jsReactTypeScriptLiveKitSpeech-to-text / TTSRubric scoring engineVercel
Next project