Grill — AI mock interview simulatoranswer by voice · text · camera~15 min per session
The interviewbefore theinterview.
It asks the follow-up you were hoping to avoid, then scores what you said and measures how you sounded — pace, pauses, fillers, pitch, energy. From the audio. Never the vibe.
Live simulation — what a session looks likenot a video — real DOM, running live
Simulated Grill session with live transcription, flagged fillers, measured delivery and an adaptive follow-up. A front-end behavioral round, question four of eight. The interviewer asks: “Tell me about a project where you missed the deadline. What slipped?” The answer comes back word by word — “So basically we underestimated the migration, um and the review cycle took longer because, like coordination between the two teams…” — with three fillers flagged where they were said: basically, um, like. Delivery, measured from the audio: 162 words per minute, longest pause 1.4 seconds, 3 fillers, 24.1 hertz of pitch variation. The session then asks its follow-up: “You said ‘we’ — what did you own, and when did you know the estimate was wrong?”
Paste the brief
A job description, your résumé, or just a topic. Pick the level and the shape.
~2 min setupTake the heat
Answer out loud or in writing, on camera if you dare. Every answer shapes the next question.
8 questions · voice or textRead the verdict
Scores grounded in your own words, plus delivery that was measured — not vibed.
receipts includedNo script.
It listens.
There is no fixed question list. Every answer decides what gets asked next — and the vaguer the answer, the harder the next question. This is the mechanic, not a feature bullet.
Every answer decides the next question — like an interviewer who was actually listening.
Measured.Not vibes.
From word-level timings in your transcription.
Every silence is on the record, to the millisecond.
Um, like, basically. Flagged where you said them.
Read from the raw audio signal itself.
JD run
Paste the job you’re actually going for. It gets probed against your résumé — gap by gap.
The full arc
The shape a real interviewer runs — opening to “any questions for us?”
Project
Grilled on something you built. Paste a write-up — or import a repo and defend the commits.
Weak spots
Re-asks the questions you scored worst on, mixed with fresh ones. The rematch you owe yourself.
Topic only
Distributed systems. React internals. Anything. Résumé deliberately ignored.
Culture only
Conflict, failure, judgement calls. No tech — just the questions that decide the offer.
Or blend freely — résumé + topic + culture in one interview that moves between all three.
Adaptive follow-ups
No fixed script. Each answer decides what gets asked next.
Voice or text
Speak or type. Speech is transcribed word by word, with timings kept.
Watch yourself back
Optional video, synced per question. See the face you make when the question lands.
Verdicts with receipts
Every score cites your own words back to you. No mystery grades.
Star the killers
Save the questions that got you and drill them again later, on your terms.
Rerun the exact brief
Repeat the same fire and watch your numbers move. Progress you can point at.
Is it actually free?
Free to start — eight questions, no card. Bring a job description and fifteen minutes; that’s the whole price of finding out.
Will it judge my tone from my words?
Never — this is the house rule. Pace and pauses come from word-level timings. Pitch and energy are read from the raw audio signal. If tonality can’t be measured, it is reported as missing — not guessed from the transcript.
What happens to my recordings?
They’re yours, scoped to your account — every session and report query is tied to your user id. Videos carry a visible expiry date, and audio is fetched on demand only when you press play.
Do I need a camera?
No. Answer by voice or text — the camera is optional, synced per question, and exists for one reason: watching yourself back is the fastest coach there is.
Will a rerun repeat the same questions?
Not unless you ask it to. Every question you’ve already faced becomes a do-not-reuse list — the same résumé never produces the same interview twice.