How to Prep for a Job Interview Using AI in 30 Minutes

Most job interview prep I've watched people do follows the same pattern. Print the job description. Re-read your CV. Write down five "tell me about yourself" answers. Rehearse in the mirror. By the time they walk into the room, they've memorized lines that sound nothing like them under pressure.

I've been running an AI loop for interview prep for about a year. Thirty minutes per pass. After two or three passes your brain stops treating the conversation as foreign territory. It's not a trick, and there's no clever prompt buried in it. It's familiarity, delivered at speed.

Here's the exact sequence.

Step 1: Drop your CV into NotebookLM. Generate the podcast. Let two synthetic voices hype you up.

Open NotebookLM. Upload your CV as a source. Add a LinkedIn export or portfolio if you have one. Generate the audio overview — once in English, once in Chinese if you work bilingual. Both.

What NotebookLM does in podcast mode is specific. Two hosts take turns narrating your career as if you're the subject of a case study. They surface accomplishments you would normally downplay. They connect dots between projects you'd never framed as a thread. A workshop you ran in Shenzhen and a training session you did in Los Angeles become "evidence of a repeatable capability" in their mouths.

You listen to this while you're making coffee. You don't have to do anything performative. But after ten minutes of two synthetic voices enthusiastically explaining why you're good at your job, something loosens. You start believing the narrative. That's the whole point of this step.

Step 2: Add the job ad and the company website. Regenerate the podcast.

Now paste the job posting into the same notebook as a second source. Pull in the company's About page. Their latest press release if there is one. Regenerate the audio overview.

This time the two hosts are building a bridge. They connect your profile to the specific role. Even if your background is a 60% match at best, the podcast finds the seams. It forces a narrative where your experience — whatever it actually is — maps onto what this company says it wants.

You don't have to agree with every bridge NotebookLM invents. Some will be a stretch. But the exercise teaches your brain the language it will need in the room. When the interviewer says "tell me about a time you led cross-functional work," you have already heard a synthetic host describe your Shenzhen workshop using that exact phrase.

Step 3: Role-play with ChatGPT in advanced voice mode. Record it. Feed the recording back into NotebookLM.

Open ChatGPT in advanced voice mode. Set the context with a system prompt along the lines of: "You're a hiring manager at [Company]. I'm interviewing for [Role]. Ask me hard behavioral and technical questions, push back when my answers are vague, and don't let me off the hook."

Then do the interview. Out loud. Full sentences. Don't type — speak.

Record the call. Your phone's voice memo app is fine. When you're done, drop the recording back into NotebookLM alongside your CV and the job ad.

Now the notebook holds three sources: your profile, the target job, and a record of you trying to bridge the two in real time. The next podcast NotebookLM generates will be informed by how you actually performed. The hosts will reference specific things you said. They will identify the gaps you stumbled on. They will rehearse the strong answers you gave, which re-encodes them.

Why the loop works, briefly

The method isn't about memorizing answers. It's about flooding your brain with the specific semantic territory of the interview — your story, the company's story, the overlap — delivered in a voice that isn't yours, before you have to perform the material in a voice that is.

Three or four loops in, something strange happens. The interview feels familiar. Not because you've rehearsed it, but because you've been marinating in the language for an evening. Your brain thinks it's been here before.

That's not a trick. That's just how exposure works. Athletes call it visualization. Actors call it tablework. For knowledge workers walking into a high-stakes conversation, this is the cheap, fast version.

What I'd skip

I've tried the version where you ask an LLM to "write me 20 interview questions for X role." It's fine, and it's fragile — you end up memorizing specific answers instead of internalizing the space. I've tried outsourcing the whole thing to a single mega-prompt in ChatGPT. That produces slop. You need the notebook structure so sources compound.

I've also seen people use AI to generate a "personal brand statement" and then read it aloud in front of a mirror. Don't. The mirror adds nothing that the NotebookLM audio doesn't already deliver, without the awkwardness.

The workflow costs nothing beyond tools most people in this market already have: a NotebookLM account (free) and a ChatGPT Plus subscription (USD 20/month). If you're between jobs, both are worth the trial.

Play smart is always cheaper than cram harder.

Connect with me on LinkedIn.