Mock Interview with AI: How It Works and How to Use It
How an AI mock interview runs from setup to report, which formats it covers, and how to get useful feedback out of each session.
Before you start: the four setup choices
Two minutes of setup decide whether the questions are generic or yours.
An AI mock interview is a simulated round in which an AI interviewer asks the questions, follows up on what you say, and scores each answer afterwards. If you want the case for practising this way and how it compares with a coach, read the complete guide to AI interview practice. This page is about the session itself: what you choose, what happens, and what comes back.
Role, seniority and round type
The role and level shape the questions; the round type shapes the interviewer. There are 12 to choose from: general, technical, behavioral, recruiter screen, system design, managerial, case study, technical assessment, leadership, panel, salary negotiation, final / executive.
Job description, company and resume
Optional, and worth the minute. With a job description the questions target its requirements; with a resume the interviewer can dig into what you actually
Question count and difficulty
5 questions on the free plan, up to 20 on Pro, at easy, medium or hard. You can add your own questions to the set.
Language and answer mode
Questions, follow-ups and feedback in any of 50+ languages. Answer by voice or by typing, and switch between them mid-session.
In the room: how the questions unfold
The interviewer has the first question planned and nothing after it.
- 1
The opening question
The interviewer introduces the round and asks the first question. There is no fixed script after this point.
- 2
Follow-ups on what you said
If an answer names a decision without the reasoning, the next question asks for the reasoning. If it is vague, you are asked for a specific example. This
- 3
Clarifying, skipping, moving on
You can ask what a question means, and you can skip one; the interviewer moves on rather than pressing. A skipped question gets no per-answer score rather
- 4
The close
After the last of the 5 (or up to 20) questions the interviewer wraps up, and the report starts generating while you wait.
One thing it will not do: feed you answers during a real interview. The tools that do are a different category — see practice tool vs live copilot.
The report: what you get back
A score per answer
Every answer is scored 0–100 for relevance, depth, clarity, with an overall mark that has to agree with the three, plus a sentence on what the score is based on.
The transcript, annotated
Each answer is shown as you gave it, with the follow-up that came after it, so you can see exactly where the interviewer felt something was missing.
Delivery numbers
From the voice transcript: filler-word count and words per minute per interview. Useful as a trend across sessions, not as a target on its own.
What to practise next
Demonstrated strengths, areas to improve, and a short list of what to focus on in the next round. You can re-answer any single question from the report
Reading the report so the next session is better
The report is only useful if it changes what you do in the next round.
- 1
Read the lowest score first
One weak dimension usually explains several weak answers. Low depth across the board means your stories lack specifics; low relevance means you are
- 2
Re-answer before you move on
The re-answer button runs the same question again. Doing it while the feedback is fresh is what makes the correction stick; reading the note a week later
- 3
Compare reports, not sessions
Reports persist. Open the previous one next to the new one and look at whether the same dimension moved. A single session is noisy; three in a row are a
- 4
Ignore the filler count until content is fixed
Fillers drop on their own once you know what you want to say. Chasing the number first produces careful, hollow answers.
Choosing between this and a human on the other side? The tool comparison groups the options by what they are for, with verified prices.
Frequently asked questions
- How long does an AI mock interview take?
- About fifteen to twenty minutes for a 5-question round including the follow-ups, and up to forty-five for a 20-question one. The report is ready a minute or two after you finish.
- Does the interviewer really follow up, or is it a fixed list?
- It follows up. The next question is generated from your answer: a claim without evidence gets asked for the evidence, a vague answer gets asked for a specific case. That is the difference between practising with a question bank and practising with an interviewer.
- What is scored, exactly?
- Three things per answer — relevance, depth, clarity — each 0–100, plus an overall score that cannot sit far above their average. Behavioral answers are judged against STAR, technical ones against problem → approach → solution. Delivery (fillers, pace) is counted from the transcript, separately from those three.
- Voice or text?
- Voice if the real interview will be spoken: the transcript shows you every trailing sentence and every "um". Text is fine for content practice or a shared office, and you can switch during a session.
- What does the free plan include?
- 2 interviews a month of 5 questions each, with the full report, no card required. Pro ($20/month or $200/year) removes the monthly cap, allows up to 20 questions a round and adds PDF reports; one-time credit packs start at $9.
- Can I re-do one question instead of the whole interview?
- Yes. Every question in the report has a re-answer option. The new attempt is scored the same way and shown beside the original, which is the fastest way to check whether a piece of feedback actually landed.