Practise with an AI that actually listens

A free AI mock interview that builds questions around your job and résumé, talks to you like a real interviewer, and tells you exactly where you're losing the room — before it costs you the offer.

In short

A free AI mock interview is a practice session where an AI plays the interviewer — it generates questions from your target job description and résumé, lets you answer out loud or in text, then scores your responses and gives instant feedback. Most tools have a free tier, but “free” ranges from unlimited forever to a single trial run. So it's worth knowing what you're actually getting.

The basics

What an AI mock interview actually is

Let's be straight about it. An AI mock interview is a rehearsal. You tell it the role you're chasing, it asks you the kind of questions that role gets asked, you answer, and it hands back feedback in seconds — no scheduling a friend, no bugging a mentor, no waiting three days for a coach to email you back.

That's the pitch everyone makes. The reality is more uneven than the marketing lets on, which we'll get to.

The good ones do three things well: they tailor the questions to your actual job description instead of throwing generic “tell me about yourself” at everyone, they let you speak your answers out loud the way you will on the day, and they give you feedback specific enough to act on. The not-so-good ones do one of those three and call it a product.


How it works

Three steps. About ten minutes to your first round.

1

Tell it the job

Paste the job description, or the URL, or upload your résumé. The more you give it, the sharper the questions. A detailed brief gets you questions about the actual skills in the posting; a vague one gets you a vague interview.

2

Pick how you want to talk

Type your answers if you're drafting and refining. Speak them if you want to rehearse delivery, pace and nerves — which is the part that actually wins or loses interviews. More on modes below.

3

Get scored, then go again

The moment you finish, you get feedback — not a vibe, a breakdown. Then you do it again, because the second run is always better than the first, and the fifth is better than the second.

The honest bit

Most “free” AI interviewers aren't built for you

Here's something the other tools won't tell you on their own landing page.

I got a real interview invite from a legit company a while back — “AI interview, optional, takes 10 minutes.” I booked it out of curiosity. The second the voice started I knew exactly what was happening, because I build AI interviewers for a living and recognised the cheap voice provider on the first sentence.

It just machine-gunned questions at me. No pauses. No reactions. No “tell me more.” Question, answer, next question, next question — the whole time I knew I was being transcribed, summarised and scored, on what, I had no idea.

That experience isn't a bug. It's what you get when a product is built to look like it works rather than to actually get you ready. Across most of the category, the pattern is the same:

  • They run the cheapest models they can. The voice and the scoring are whatever costs least — robotic delivery, shallow feedback — because the model bill is the first thing they cut.
  • They don't care about the experience. It's functional, not useful — a box that technically conducts an interview and ticks a feature off a roadmap. Whether you walk away more prepared was never the point.
  • They aren't built for the person practising. They're built for the company buying, or the demo that closes the funding round. The candidate is something to process, not someone to serve.
  • They don't give you real free practice. “Free” is a single trial or a teaser, engineered to push you to pay — not to let you rehearse, again and again, until you're genuinely ready.

AI interviewing is going to become completely normal — that ship has sailed. But talking to an AI isn't the same as talking to a human, and most candidates prep for one and walk into the other. The gap isn't the technology. It's that nobody's building the practice for the thing you'll actually face.

That's the whole reason withLyra exists.

Why us

Why withLyra is the best free AI mock interview

We made the calls most free tools won't, because cheaping out on the parts that matter is how you end up with a question generator.

The best voice model, on the free tier

The most natural conversational voice available — not the hold-music robot most free tools ship.

Seven aspects of feedback, not a single score

Feedback across seven aspects — so you know exactly what to fix, not just a number you can't act on.

Tailored to you, not the same questions for everyone

Built around your job description and background — the questions this role actually asks.

Practice mode for the reps in between

Write your own questions or let withLyra generate them, drill them anytime, then run your bank as a full mock.

A record that grows with your career

Every session is saved, so you can watch yourself improve and catch the weakness that keeps resurfacing.

Three interviewers, three styles

Practise with Sophia, Vince or Daniel — different personas, different energy. Real interviewers vary; your practice should too.

Start free — no card neededNo credit card required.

Pick your mode

Text, voice, or conversation

Different modes, different jobs.

Text

Best for the early drafting stage, when you're still working out what you want to say and want to refine the words.

Voice

Best for delivery: your pace, your tone, the filler words, the pauses. This is where most interviews are actually won or lost — the mode people skip because it's uncomfortable. Do it anyway.

Conversation

The most realistic: a back-and-forth that reacts to your answers instead of marching through a script. The closest thing to the real room.

Our honest advice: draft in text if you must, but spend your reps in voice and conversation. You don't get to type your answers on the day.

Practice mode

Practise on your terms — your questions, or ours

Full mock interviews are the main event. Practice mode is for the reps in between.

Write your own questions

Got a question you keep fumbling? Add it to your bank and drill it as many times as you need — on your schedule, not a timed session.

Or let withLyra generate them

Don't know what you'll be asked? Tell us the role and we'll generate a tailored set of likely questions for you to work through.

Build a bank, then run it as a mock

Like a question? Save it. When you're ready, pick the ones you want and turn your bank into a full mock interview in one tap.

Your question bankGenerate for me
Behavioural

Tell me about a time you disagreed with your manager. How did you handle it?

Technical

Walk me through how you'd design a rate limiter for a public API.

Write your own question

No black box

What the feedback actually looks at

This is the part nobody likes to own up to: when an AI scores you, you usually have no idea what the rubric is. Being graded by a black box on your tone and your filler words, with zero visibility into the criteria, is the genuinely uncomfortable bit of this whole shift.

So we show ours. Your feedback spans seven aspects of how you came across — the substance of your answers, how clearly they were structured, how well they matched the question, your delivery, and more. Each one is something you can actually work on before the next round. No mystery number. No “trust us.”

Relevance
Structure
Clarity
Job fit
Seniority
Confidence
Ownership

A fair question

How is this different from just using ChatGPT?

Fair question — you can ask a chatbot to quiz you.

But a text chat with a general assistant isn't an interview. It doesn't talk back in real time, it doesn't hold a consistent interviewer persona, it won't score your delivery because it can't hear you, and it won't keep a record of how you're improving across sessions. You're doing the work of being your own interviewer, scorer and progress tracker — which rather defeats the point of practising.

A purpose-built mock interview runs the room so you can focus on being the candidate.


Straight answer

Are AI mock interviews accurate?

Mostly — with a caveat worth stating plainly.

The questions are genuinely good when you feed in a real job description — the more detail you give, the closer they map to what you'll be asked. The feedback is grounded in interview best practice (think STAR-method structure), and it's consistent in a way a nervous human panel often isn't.

The caveat: AI can occasionally get something wrong or overstate a point. Treat the feedback as a sharp, tireless practice partner — not a hiring oracle. The point isn't a perfect prediction of your real interview. The point is reps, a clear read on your weak spots, and walking in less rattled than you would have otherwise.


Make it count

How to get the most out of it

Use a real job description

Specific in, specific out. This is the single biggest lever.

Answer out loud

Reading answers silently builds false confidence. Speaking them builds the real thing.

Lean on STAR for behavioural questions

Situation, Task, Action, Result. It stops you rambling — and it's what the feedback is checking for anyway.

Do more than one round

First run shakes out the nerves. The improvement shows up on runs two through five.

Watch your history

If the same weakness keeps surfacing across sessions, that's your homework. The record is there to be used.


FAQ

Questions, answered straight

A practice job interview where an AI plays the interviewer. It asks role-specific questions, you answer by voice or text, and it gives you instant feedback on how you did.
Often, yes — but “free” varies. Some tools are free forever, some give you one trial round, some hand out a few credits a month. withLyra's free tier runs on the best voice model and doesn't make you hand over a card to start.
You give it your target job (description, URL or résumé), it generates tailored questions, you answer, and it scores your responses across several aspects with feedback you can act on straight away.
The questions are strong when you supply a detailed job description, and the feedback is grounded in interview best practice. AI can occasionally get something wrong, so treat it as a practice partner rather than a guaranteed prediction of your real interview.
A general chatbot won't talk back in real time, hold an interviewer persona, score your spoken delivery, or track your progress over time. A purpose-built tool runs the interview so you can focus on answering.
Yes. Because the interview is built from your specific job description and background, it adapts to the role you're chasing rather than a fixed list of questions.
Text for drafting what you want to say; voice for rehearsing delivery, pace and nerves; conversation for the most realistic, reactive back-and-forth. Spend most of your reps in voice and conversation.
Not always — a lot of real AI interviews are blunt, machine-gun question generators with no flow. Practising with something more realistic means the real thing, good or rough, won't catch you off guard.
More than once. The first round settles your nerves; the real gains show up over several rounds, especially in the week before a big interview.
Sessions are saved so you can review them and track your progress over time.
Get started

Stop rehearsing in your head. Start your free mock interview.

Start your free mock interview →

No credit card required.