Screen for what you actually need.
Bring your own rubric.
A voice AI runs first-rounds against your per-criterion rubric, not a generic screening template. You see how candidates actually reason — before you spend an onsite slot finding out.
Built on LiveKit · LL144-ready · 3 free, no card
What it is
Vettika for hiring managers is a voice-AI first-round interviewer that scores candidates per-criterion against the rubric you write — not a generic screening template. Engineering managers and department heads typically lose 4 to 6 onsite slots per role to candidates the recruiter screen forwarded on resume keywords. Vettika cuts that to one or zero: every candidate ships with a per-criterion score and a full transcript so you read the actual reasoning before committing the loop. First 3 interviews free with no card. After that, a 10-pack at $19 ($1.90 each), a 50-pack at $79 ($1.58 each), or Pro at $49/mo for 30 interviews plus $1.50 per extra. No-shows never count.
The hiring-manager screening problem
Recruiter screens for the wrong signal. You pay for it later.
Was
Recruiter screen filtered for resume keywords, not the skill you actually need
Now
You write the rubric. Agent scores per-criterion. Resume keywords are no longer the funnel.
Was
Sit through onsite #4 just to learn the candidate cannot reason through a tradeoff
Now
See how they reason in the 12-minute transcript. Pass on weak reasoners before booking the loop.
Was
Recruiter notes are vibes; you cannot tell why a candidate was forwarded
Now
Every candidate ships with a per-criterion score and the actual transcript. No vibes.
How a hiring manager uses it end to end
From rubric to onsite decision in 4 minutes per candidate.
Write the rubric you would have given the recruiter
List the 4-6 things you actually want to test — systems thinking, debugging, ownership, async writing, prior infra scale. One sentence per criterion.
Set the questions and the bar
Pick 4-6 questions that surface those criteria. Set a pass-bar per criterion. The agent will press on weak answers and not let candidates wave them away.
Hand the link to the recruiter (or post it yourself)
Recruiter routes applicants through your link. You get the scored transcripts. Recruiter still owns the funnel; you own the bar.
Read 5 transcripts, book 3 onsites
Each transcript takes 4 minutes to read. The per-criterion score tells you where to look. Onsite slots only go to candidates who cleared your bar on tape.
Get back 5 onsite slots per role.
A typical engineering loop is 4 to 5 interviewers × 1 hour each. Killing 5 mis-fit onsites per role saves your panel roughly 20-25 engineer-hours — and saves the candidates a trip they should not have taken. The 12-minute screen on the front end pays for itself the first time you catch a weak reasoner before the loop.
Pricing for hiring managers
Cheaper than one wasted onsite.
First 3 interviews free, no card. After that, a 10-pack at $19 ($1.90 each) if you hire 1-2 roles a quarter, or Pro at $49/mo for 30 interviews ($1.50/extra) if you are running a team-wide hiring sprint. No-shows never count against your cap.
No card. No subscription.
Per-criterion rubric you write
Full transcript, not just score
Audit log per session
FAQ
Questions hiring managers actually ask.
Can I really bring my own rubric, or is this just templated scoring?+
Your rubric, your criteria, your bar. You list the 4-6 dimensions that matter for the role — for an EM hiring a senior backend engineer that might be (1) systems design depth, (2) debugging methodology, (3) team-impact ownership, (4) async written communication. The agent scores each candidate per-criterion against the bar you set. No generic templates unless you want them.
I do not trust AI to score technical depth — what is the failure mode?+
Honest answer: the agent is excellent at extracting what the candidate said and reasonable at scoring whether the answer demonstrates the criterion. It will not catch a subtly wrong distributed-systems claim that requires expert review. Use the score to filter; use the transcript to verify. Read the actual exchange before you commit an onsite slot. We make the transcript the source of truth, not the score.
Will the recruiter feel bypassed?+
No — they get the same shortlist they would have built, with better signal per candidate. Most hiring manager partnerships we have seen route Vettika as a layer between recruiter screen and onsite, not as a replacement for the recruiter relationship. The recruiter still sources, still tags candidates, still owns the funnel.
How does this work for non-engineering roles?+
Same shape, different criteria. A product manager hiring a senior PM might rubric on (1) prioritization framework, (2) customer-discovery technique, (3) cross-functional negotiation, (4) data literacy. A design lead hiring a senior designer might rubric on craft, critique, and systems thinking. The product is rubric-agnostic; it scores whatever you write.
Can I see the audit trail for compliance?+
Yes. Per-session audit log: rubric used, questions asked, answers transcribed, per-criterion scores, time stamps. LL144 bias-audit format. EU AI Act risk-classification documentation published. Exportable per session and per role.
Related
Keep reading
Your bar, on every first-round.
Write the rubric once. Score every candidate against it. First 3 free.