Find the strong engineers in a pile of polished résumés.
Vesna turns the job into must-haves, checks every applicant against them and asks candidates about missing technical details, so your engineers interview people who have shown the work.
One applicant, five must-haves. Select a must-have to see what it rests on.
The problem
Inbound is a pile nobody can read.
A role gets 500 polished résumés. A team reads the first 150 and stops. The strong people further down are never opened.
The process
How Vesna fits your hiring process.
Vesna checks each résumé against the must-haves your team agreed, and emails the candidate when the résumé doesn’t cover one. Only a person rejects.
Step 1
Agree on what counts.
Vesna turns your job description into must-haves and says how each one gets checked. Your team edits them before any applicant is read.
Names a backend service or feature they built in TypeScript or Node.js and ran in production for real users or other systems. Listing TypeScript or Node.js as a skill does not show this.
Vesna asks the candidate. A résumé line can claim it; only the candidate can ground it.
Vesna prepares interview questions. It only holds up under a question asked in the interview.
Vesna checks the trace or asks for it. It leaves a trace someone else can check.
Names a product feature they built across both a React or Next.js front end and the backend behind it. Front-end-only or back-end-only work does not show this.
Vesna asks the candidate. A résumé line can claim it; only the candidate can ground it.
Vesna prepares interview questions. It only holds up under a question asked in the interview.
Vesna checks the trace or asks for it. It leaves a trace someone else can check.
Names how they kept writes into another company’s system correct when calls timed out, repeated or arrived late, for example idempotency keys or checking the other system before retrying.
Vesna asks the candidate. A résumé line can claim it; only the candidate can ground it.
Vesna prepares interview questions. It only holds up under a question asked in the interview.
Vesna checks the trace or asks for it. It leaves a trace someone else can check.
Names one real production failure they traced to its cause. A general account of how they debug does not show this.
Vesna asks the candidate. A résumé line can claim it; only the candidate can ground it.
Vesna prepares interview questions. It only holds up under a question asked in the interview.
Vesna checks the trace or asks for it. It leaves a trace someone else can check.
Names real engineering work they handed to a coding agent and something they changed when the agent got stuck or got it wrong. Saying they use AI tools does not show this.
Vesna asks the candidate. A résumé line can claim it; only the candidate can ground it.
Vesna prepares interview questions. It only holds up under a question asked in the interview.
Vesna checks the trace or asks for it. It leaves a trace someone else can check.
Step 2
Connect your applicant flow.
Applicants come in from Greenhouse, Ashby, Lever or Workable, by email, through the API or as an upload.
Step 3
Read every applicant against the role.
One reading per must-have, with the résumé lines it rests on. No overall score.
Shown: the résumé or an answer proves it. Claimed: the résumé says it, without detail. No evidence yet: Vesna asks the candidate.
Step 4
Ask before you rule anyone out.
When a résumé doesn’t mention a must-have, Vesna emails the candidate one question about it and tells you whether the answer meets it.
Chrystian Schutz
Ran a multi-agent C#-to-Java migration pipeline.
Claimed. It doesn’t say what he changed when it went wrong.
Hooli via Vesna
Hooli has one quick question about your application
“When the generator silently improved legacy logic instead of replicating it: what exactly did you change, and how did you know that change was the fix?”
Answer the questionIt takes a couple of minutes. This is not a test.
What exactly did you change to fix that
Thank you for this thoughtful question. As a results-driven engineer I always strive to uphold the highest standards. I split the pipeline so a second agent only compares the Java output with the C# build. This reflects my deep commitment to robust engineering. Mismatched totals went from 31 modules to none on the next run.
It asks only when the answer can move the candidate.
Five must-haves, ranked by how far an emailed answer could move him. Bar lengths are illustrative.
Ranked 1 of 5
Changed how they work with coding agents
His résumé names a real pipeline and how it failed, so a precise question is possible. A clear answer settles this must-have.
“When the generator silently improved legacy logic instead of replicating it: what exactly did you change, and how did you know that change was the fix?”
Ranked 2 of 5
Shipped features across a React front end and its backend
The claim is vague, so the question has to be broader and one answer is less likely to settle it. Vesna asks it after the first answer is in.
Ranked 3 of 5
Owned a TypeScript service in production
His résumé already proves it. No answer would add anything, so Vesna doesn’t take his time.
Ranked 4 of 5
Kept writes correct in systems they did not control
Your team set this must-have to be checked in the interview. An emailed answer wouldn’t count as proof, so Vesna doesn’t send one. The question goes into the interview brief.
Ranked 5 of 5
Found the cause of a hard production failure
Your team set this must-have to be checked in the interview. An emailed answer wouldn’t count as proof, so Vesna doesn’t send one. The question goes into the interview brief.
Step 5
Hand over the shortlist.
The hiring manager gets each shortlisted person with why they advanced and what to ask, and replies on the same page.
Mei Lin Tan
Senior Backend Engineer, Node.js and TypeScript
“I did not review it until it was finished, and the Kafka consumer already wrote each batch to Postgres in a single transaction…”
What risk did she identify? What did she change? What improved?
“Prepare my interview with Mei. Skip what we already know; give me the gaps and the evidence.”
At the scale of a real pool
Everyone is read. Some are asked. No one is rejected by a machine.
Open a role to inbound without dreading the pile. Vesna advances the people who show the work and leaves everyone else on hold for a person.
The evidence trail
Every step is on the record.
What Vesna read, what it concluded and when. You can always answer “why?”, to a hiring manager or to a candidate.
The PDF from your applicant flow
Stored unchanged
System
The résumé and the five “what counts as shown” texts
1 shown, 2 claimed, 2 for the interview, each with its résumé line
Vesna
The two claimed must-haves
Coding agents first: his pipeline line names a failure but not the fix
Vesna
The drafted question
One email to the candidate
Vesna
No answer after two days
One reminder
Vesna
Typed by the candidate, 71 words
Stored word for word
Candidate
His answer and the “what counts as shown” text
Shown: names what he changed and how he checked it
Vesna
All must-haves Vesna can check are shown
On the shortlist, evidence attached
Vesna. Only a person can reject.
What you get back
Give engineering a shortlist with the groundwork done.
Your engineers see what each candidate has shown before the interview, so the interview goes on what is still unknown.
How much of the pile nobody reads
Your numbers. It assumes your team reads in the order applications arrive, so strong people are spread through the pile.
With VesnaEvery applicant is read against the must-haves, the 500th as closely as the first.
Fewer engineering hours in interviews
One engineer’s week of first interviews. Illustrative.
No second shortlist
From open role to first interviews. Illustrative lengths.
Compare
What happens when the résumé isn’t enough?
How Vesna and other screening tools handle a résumé that doesn’t answer a must-have.
| Workflow | Workable Agent | Vesna |
|---|---|---|
| When the résumé is vague | Asks for missing details and updates evaluations from the replies | Asks about one requirement, in that requirement’s terms: “Describe one comment you left that you later saw reflected in how that engineer wrote code on their own.” Then shows you whether the answer meets the must-have your team agreed. |
| How a candidate is judged | Criteria-based evaluation with reasoning and a normalized score | No overall score. One reading per must-have, with the résumé line or answer it rests on. |
| Prepare the next step | Configurable pipeline actions | For each candidate: what is shown for each must-have, and what to ask in the engineering interview. |
| Keep your workflow | Inside Workable | Alongside your existing ATS |
From Workable’s product documentation. Reviewed 25 October 2026.
| Workflow | Ashby | Vesna |
|---|---|---|
| When the résumé is vague | Application-form answers can be assessed against criteria | Asks about one requirement, in that requirement’s terms: “Describe one comment you left that you later saw reflected in how that engineer wrote code on their own.” Then shows you whether the answer meets the must-have your team agreed. |
| How a candidate is judged | Custom criteria with rationale and source citations | No overall score. One reading per must-have, with the résumé line or answer it rests on. |
| Prepare the next step | Application review and filtering | For each candidate: what is shown for each must-have, and what to ask in the engineering interview. |
| Keep your workflow | Inside Ashby | Alongside your existing ATS |
From Ashby’s product documentation. Reviewed 25 October 2026.
| Workflow | Greenhouse Real Talent | Vesna |
|---|---|---|
| When the résumé is vague | The documented focus is matching and verification | Asks about one requirement, in that requirement’s terms: “Describe one comment you left that you later saw reflected in how that engineer wrote code on their own.” Then shows you whether the answer meets the must-have your team agreed. |
| How a candidate is judged | Candidate matching, fraud detection and identity verification | No overall score. One reading per must-have, with the résumé line or answer it rests on. |
| Prepare the next step | Recruiter review within Greenhouse | For each candidate: what is shown for each must-have, and what to ask in the engineering interview. |
| Keep your workflow | Inside Greenhouse | Alongside your existing ATS |
From Greenhouse’s product documentation. Reviewed 25 October 2026.
| Workflow | Hirevire | Vesna |
|---|---|---|
| When the résumé is vague | Follow-up questions after response review | Asks about one requirement, in that requirement’s terms: “Describe one comment you left that you later saw reflected in how that engineer wrote code on their own.” Then shows you whether the answer meets the must-have your team agreed. |
| How a candidate is judged | Video, audio, text responses and AI scorecards | No overall score. One reading per must-have, with the résumé line or answer it rests on. |
| Prepare the next step | Team review, sharing and configurable stages | For each candidate: what is shown for each must-have, and what to ask in the engineering interview. |
| Keep your workflow | A separate screening platform | Alongside your existing ATS |
From Hirevire’s product documentation. Reviewed 25 October 2026.
Questions a hiring team asks first.
Does Vesna reject anyone on its own?
No. Vesna advances candidates or leaves them on hold. Only a person rejects.
What does the candidate see?
An email with one plain question and a page to answer it, typed or dictated.
Do you reject résumés written with AI?
No. A polished résumé still needs detail behind its claims, and a sparse one gets a question instead of a rejection.
Who can see our applicants’ data?
Only your organisation: each company’s data is kept apart in the database. Résumés are read through Anthropic’s commercial API. How Vesna protects hiring data
What is recorded?
Every action and every AI call is logged with who, when and the evidence, and can’t be edited. You can export it.
Which ATS do you work with?
Greenhouse, Ashby, Lever and Workable. Email, API and upload work too.
How long does setup take?
About a day. You send a job link, edit the must-haves Vesna proposes, and it starts reading applicants.
Try Vesna on one of your open roles.
Send a public job link. Within one working day you get the proposed must-haves and the questions Vesna would ask.