// Reverse mode
Switch seats.
You play the interviewer. The AI plays the best candidate for the job: the one the job description is looking for. Watch a strong answer happen in real time, ask the follow-ups you dread, and see how they handle them.
// 01 — How it works
Learn by watching the other side
- 01
Pick the chair
Choose a round and a problem: from the question bank, from your interview plan, or, for system design, one you bring. The candidate is built from the job: its level, stack, and domain.
- 02
Run the interview
Ask the question. The AI candidate works through it out loud, the way the best candidate for the role would, and asks you questions back.
- 03
Steer, then debrief
Interrupt, push back, change a requirement. Afterwards, get a breakdown of the moves the candidate made and why they worked.
- 00:12You
“Let’s design a URL shortener.”
- 00:20CandidateClarifying
“Before I draw anything: is this public, and roughly how many redirects a day are we expecting?”
- 00:31You
“Public. Say a hundred million redirects a day.”
- 00:44CandidateEstimating
“That’s around twelve hundred reads a second on average, so the redirect path should be cache-first. Writes are rare by comparison. I’ll keep them simple and spend my time on reads.”
- 01:10CandidateAsking you
“Is it acceptable if a new link takes a second or two to work everywhere, or does it need to be instant?”
// 02 — What to watch for
The moves that separate a strong candidate from a correct one
01 —
Clarify first
Requirements and constraints before a single box is drawn.
02 —
Show the math
Estimates said out loud, and used to justify choices.
03 —
Name the tradeoff
What each decision costs, not just what it buys.
04 —
Ask good questions
The questions a strong candidate asks the interviewer.
05 —
Recover from a wrong turn
Noticing a flaw in its own approach and correcting course out loud.
06 —
Manage the clock
Knowing when to go deep and when to move on.
// 03 — Why it works
Most people have never seen a good interview from the outside
Reading a model answer shows you where to end up. It doesn’t show you how to get there: the clarifying question asked in the first minute, the estimate that rules out half the options, the moment a candidate says “actually, that won’t hold up” and recovers.
Sitting in the interviewer’s chair also changes how you hear your own answers. Once you’ve tried to evaluate a vague response, you stop giving them. Use Switch seats before a round to see the shape of a strong answer, then run the same round yourself.
// 04 — Available rounds
Switch seats works with
- Intro & recruiter screen> so, walk me through your background.
- Technical interview> what actually happens when that query runs?
- System design> what breaks first at 10× traffic?
- Leadership & behavioral> and what did you do when they disagreed?
- Product sense> how would you know this feature worked?
- Salary negotiation> this is already at the top of our band.
There’s no difficulty setting and your CV isn’t used: the AI candidate always aims to be the best candidate for the job. Its background is fictional, built from the job description. Switch seats isn’t available for live coding or take-home rounds. A Switch seats session costs the same credits as the round it’s based on. See pricing.
// Player 1
Get ready for hard mode.
Paste the job you want. Run the loop until the real one feels like a rematch.
20 free credits · No card required