Voice-first candidate screening
Every applicant sounds brilliant on paper. Find out who actually is.
Your AI interviewer calls every candidate, drills into what they actually built, catches the ones reading answers off a second screen, and hands you a ranked shortlist. 100 applicants screened before you finish your coffee.
- ~15 min per candidate
- 1–10 weighted score
- Zero founder hours
The problem
Applications stopped being a filter.
Every candidate now writes with AI. Portfolios are polished, the stack vocabulary is correct, the case studies are formatted. Screen a hundred of them and they all read like an 8/10 — because the things that separate a real operator from a confident writer never appear in a CV.
No run volumes
Nobody writes down how much traffic their build actually handled.
No running cost
Nobody lists what their system costs per month to keep alive.
Nothing ever broke
Nobody documents the failure they had to fix at 2am. That's where the signal is.
Reading 100 applications and running 100 intro calls is roughly 40 hours of your week. And at the end of it you still can't tell who's real.
How it works
Four steps. You show up at the end.
01
Send the invite.
Each candidate gets a unique, single-use link. It expires after one session, so nobody shares it, nobody runs it twice, nobody sends a friend.
02
The agent interviews them.
A 15-minute structured voice conversation. It follows their own claims instead of a fixed script — asks what broke, what it cost, what they'd do differently, and demands names, numbers and versions.
03
It scores while they talk.
Response timing, answer specificity, depth under pressure, and willingness to push back all feed a weighted score. Generic fluency gets marked down, not up.
04
You read a ranked list.
One row per candidate, ranked 1–10, with the transcript and audio attached. Listen only to the ones worth listening to.
The hard part
A voice bot that just asks questions gets beaten in five minutes.
Anyone can run ChatGPT on a second screen while they talk to you. So this agent is built around the things a second screen can't rescue.
Response latency analysis
A candidate typing a question into another window answers a beat too late, too consistently. We measure the gap between question and speech across the whole conversation and flag the pattern, not the one-off pause.
Claim-chained probing
The agent doesn't run a question list. It takes what the candidate just claimed and drills three levels into it. Borrowed answers collapse at level two.
Failure-first questioning
It asks what broke, not what they built. Describing a system you didn't run is easy. Describing how it failed is not.
Rate discovery, not negotiation
It finds their actual floor and states clearly that nothing discussed is an offer. You keep every bit of negotiating room.
The output
One screen. Ranked top to bottom.
Scores, risk flags, asked rates and full transcripts land in a single dashboard — or straight in your recruiter's inbox. No spreadsheet wrangling.
104 SCREENED · 6 SHORTLISTED · UPDATED 04:12
RANKED OUTPUT
| RANK | CANDIDATE | SCORE | DEPTH | AI-ASSIST RISK | RATE (ASKED) | STATUS |
|---|---|---|---|---|---|---|
| 01 | M. Okonkwo | 9.2 | HIGH | LOW | $34/HR | TRANSCRIPT READY |
| 02 | S. Bergström | 8.7 | HIGH | LOW | $41/HR | TRANSCRIPT READY |
| 03 | A. Villanueva | 8.1 | MED | LOW | $12/HR | TRANSCRIPT READY |
| 04 | D. Kaur | 6.4 | MED | MED | $27/HR | TRANSCRIPT READY |
| 05 | J. Whitfield | 4.9 | LOW | HIGH | $38/HR | REVIEW |
| 06 | T. Nakamura | 3.2 | LOW | HIGH | $22/HR | REVIEW |
- Weighted component scoring
- Audio + full transcript per candidate
- Automated email delivery to your recruiter
Capabilities
Built for high-volume hiring.
Role-specific question banks
Same engine, different ammunition. Swap the probe set for developers, editors, ops leads or setters in minutes.
Single-use secure invites
Tokenised links, one session each, hard time cap per call so your usage bill can't run away.
Configurable scoring weights
Decide what actually matters for the seat. Weight depth over polish, or ops experience over credentials.
Bonuses, not gates
Credentials count toward a score, never block a candidate. The best operators rarely have the expected background.
Multilingual screening
Run the same structured interview in English, Spanish, German or Tagalog without rebuilding it.
Human always decides
The agent ranks and evidences. Hiring calls stay with your team, with the transcript in front of them.
Use cases
Where it earns its keep.
Technical & automation hires
Separate people who have shipped production automations from people who have watched tutorials about them. Probes run volumes, cost, monitoring, and what failed.
Creative & editing roles
Turnaround times, retention pacing, revision handling, actual software depth. Portfolio reels don't tell you who hits a deadline.
Offshore & high-volume pipelines
Built for the 100+ applicant reality of OnlineJobs.ph, Upwork and LinkedIn floods. Every candidate gets the same rigorous screen at the same cost.
The math
100 candidates, two ways.
| Manual screening | Recruiter AI Test | |
|---|---|---|
| Time to screen 100 | ~40 hours of your week | One evening, unattended |
| Consistency | Depends on which call it was and how tired you were | Identical structure and scoring, every candidate |
| AI-assisted answers | Usually invisible | Flagged with timing evidence |
| Depth of questioning | Fades after candidate 20 | Same on candidate 100 as candidate 1 |
| Output | Notes you'll never re-read | Ranked scorecard, transcript, audio |
| Cost per candidate | Your hourly rate | Fraction of one hour |
Get started
Scoped to your roles, not to a plan tier.
Every screen is built around the specific seat you're filling, so we quote it after a conversation — never off a pricing table.
Single Role
One role, one question bank, one hiring round. Agent build, scoring weights, invite links and ranked output delivered ready to send.
Contact Us Directly- Most common
Hiring Engine
Multiple roles, reusable question banks, recruiter dashboard, automated email delivery, and ongoing tuning as you learn what predicts performance.
Contact Us Directly Embedded
We run it as part of your recruiting operation. Custom scoring models, integrations into your existing HR stack, and support for your recruiters.
Contact Us Directly
- Candidates told clearly they're speaking with AI
- Human review on every hiring decision
- Transcripts and scoring logic fully auditable
- GDPR-conscious data handling and retention limits
FAQ
Straight answers.
Stop reading applications. Start listening to the good ones.
Tell us the role you're hiring for and how many applicants are sitting in your inbox. We'll show you exactly what the output looks like.