vesna

AI polished your pool. Your rubric got distrusted. Your decisions got harder to defend.

Vesna is an audit-ready hiring copilot — it reads candidates for substance, asks when paper can't tell, and leaves a record you can defend.

For recruiters and hiring leads who have to triage an AI-polished pool on substance — and stand behind every call.

No auto-rejects — every decision routes to a human

Book a 15-min conversation → See a real run →

Screening broke quietly

Most candidates now use AI to write their applications. Almost every résumé looks strong — fluent, full of the right keywords, confidently specific. Three things followed, and you've felt all of them.

2in 3
candidates use AI somewhere in their application
CNBC, 2025 ↗
1
keyword filter that now floats the fluent and buries the real
0
good ways left to defend "why did we pass on them?"
Everyone looks strong
"Polished" stopped meaning "qualified" — but it still wins the keyword screen. The honest, plainly-written candidate gets buried; the one who sounds great but can't back it up rises.
Rubric distrusted
When everyone matches the must-haves on paper, the rubric stops telling candidates apart — and your team quietly stops trusting it.
Harder to defend
"Why did we pass?" is a harder question than it was two years ago — and new rules on AI in hiring (like the EU AI Act) are coming.

What Vesna actually does

What makes Vesna work isn't a chat box. It's a fixed, repeatable way to gather evidence and record decisions — three narrow jobs, done in order.

sorts on what can be checked every step, captured INPUT A résumé lands — likely AI-polished, like most of the pool. 01 · POLISH-RESISTANT READ Read for substance, not polish — a graded signal, never an auto-reject. CLEAR ON PAPER → straight to the queue SAME ON PAPER 02 · ask one follow-up question QUEUE · RANKED BY FIT × ENGAGEMENT Strong + engaged rise; ghosts and hollow-polish sink. 03 · APPEND-ONLY AUDIT TRAIL Every step on the record — "why did we pass?" is one screen.
Three narrow jobs, in order. The only fork is sorting: candidates who are clear on paper go straight to the human; candidates who look the same on paper get one follow-up question first. Nothing is auto-rejected.

We don't just claim it can't be gamed — we measured it

Most vendors say "trust the AI." We didn't want you to. So we built a test that runs the real product against candidates trying to game it and recruiters cutting corners — every combination — and graded each result against the known-right answer.

48
candidate-and-recruiter combinations, run start to finish
0
cases Vesna sorted wrong, in the test
1
risk left: a human overriding a correct flag — which the record catches

A candidate trying to game it couldn't fake their way through. Across every combination we tested, Vesna sorted every case correctly — and we can run the whole test again any time. It's not a one-off demo.

See exactly how one role runs → How the data & the AI work →

Who it's for

Head of Talent / Recruiting

Triage a polished pool without false-flagging honest people, and defend every decision. For talent leaders →

CTO / Engineering Manager

Stop interviewing candidates who sound great but can't back it up; get a clear read on who actually fits your must-haves. For engineering leaders →

Where we are

In early access. Looking for the first three design partners — a focused, two-week diagnostic on your real engineering roles. See what the pilot involves →

Book a 15-min conversation → Founder on LinkedIn ↗