“Walk into the interview having already lived it.”
Upload a CV and a job description. SimuPrep builds a realistic AI interviewer that adapts to every answer, challenges the experience on the page and shows the candidate exactly where to improve afterwards.
For candidates preparing for a specific role, from first job to senior hire.

- 6interviewer personas, each with voice and video
- 1:1follow-up questions built from the last answer
- 0live judgement shown during the interview
The product is in active development with a small group of testers. We can walk you through the working build on a call.
The parts people notice.
- 01
It follows up on what you actually said
A real interviewer changes the next question based on your last answer. So does SimuPrep. Vague claims get probed, specifics get credit.
- 02
Built from your evidence
The CV and job description are parsed into an evidence map, so questions target the gap between what the role needs and what you have shown.
- 03
A readiness score with a plan
After each session: a score, your strongest and weakest answers and what to practise next. Progress is tracked per target role.
- 04
Private by design
Your camera is never analysed for confidence, emotion or appearance. Documents stay private and can be deleted at any time.
Three screens, no manual.
- 01
Upload
Add your CV and the job description. SimuPrep maps your evidence against the role.
- 02
Simulate
Enter a live interview with an AI that adapts to every answer and remembers what you said.
- 03
Improve.
Get a readiness score, your strongest and weakest answers and what to practise next.
The thing it does that nothing else does.
Shown live on the left, as it appears in the product on the right.

Under the bonnet.
- Next.js 16
- React and TypeScript
- PostgreSQL
- Speech and language models
- Video interviewer loops
- CV parsing (PDF and Word)
- Radix UI and Motion
- Usage-metered plans
Adaptive questioning over a fixed script
Question banks are easy and feel fake. We built the interviewer around the transcript so far, which is harder to engineer and much closer to the real thing.
Interviewers that look up when you speak
Each persona has pre-rendered talking and listening loops that swap on speech events. It feels present without the cost of generating video live.
No surveillance features, on purpose
We chose not to score eye contact or expression. The feedback is about the answers, which is what candidates can actually change.
Minutes as the unit of value
Plans meter avatar and voice minutes, so pricing tracks real cost and a free tier stays affordable to run.
What we learned building it, now yours.
- Business tools and automation
Conversational AI that stays in bounds
The same prompt structure, evaluation and guard rails run the assistant on this website.
See the service - Web development
Document intake that just works
PDF and Word parsing into structured fields, reused for CV, invoice and form uploads in client projects.
See the service
Pick a time. We’ll arrive prepared.
Times are shown in your time zone. You get a confirmation and calendar invite straight away, and a link to move it if life happens.


