Mock Interviews with AI: Practice with Claude or ChatGPT

Most professionals prepare for interviews by asking ChatGPT to "give me 10 common behavioral questions for a Project Manager." The AI spits out generic, textbook questions, and the user reads them and drafts answers in their head. This method is ineffective—it doesn't practice the physical act of speaking, and it fails to prepare you for a tough interviewer who probes your metrics.

With modern LLMs like ChatGPT and Gemini, you can turn your phone into a strict, realistic interviewer. By combining a roleplaying prompt with real-time mobile voice modes, you can run realistic, spoken mock interviews that actually build conversational muscle memory.

Why Standard AI Interview Prep Fails

If you don't control the AI's behavior, it will quickly break the interview simulation. Users typically face three major frustrations:

• The Script Reader: The AI prints the entire script (both the questions and your mock answers) in a single turn instead of asking them one by one.
• The Soft Critique: Chatbots are default-programmed to be polite. When you give a vague answer, they say "Excellent!" instead of calling out your lack of metrics.
• Typing Lag: Typing out responses gives your brain minutes to formulate a perfect sentence. In a real interview, you have seconds to respond verbally.

To solve this, we use a two-step prompt system: first, a **System Mock Interview Prompt** that locks the AI into a strict step-by-step roleplay, and second, a **STAR Grading Prompt** that forces a critical scorecard output once the interview is complete.

Step 1: Build Your System Prompt

Use the generator below to build a customized system prompt based on your target role, experience level, and preferred interview style. Copy this prompt and paste it into ChatGPT or Claude to begin the session.

Step 2: Go Hands-Free with Voice Mode

Typing your answers defeats the purpose of practice. To get the maximum benefit, use the voice feature of your AI app to simulate a real phone or video call:

1. Paste the prompt: Open the ChatGPT or Gemini app on your phone. Paste your generated system prompt into a new chat and press send.
2. Launch Voice Mode:
   • In ChatGPT: Tap the headphone icon next to the text field to launch Advanced Voice Mode.
   • In Gemini: Tap the wave/sparkle icon next to the text field to launch Gemini Live.
3. Speak naturally: The AI will introduce itself verbally. Respond out loud. Do not touch your phone—let the conversational flow proceed naturally. The AI will listen, wait for you to finish, and ask its follow-ups.

Step 3: Grade Your Answers Under the STAR Method

Once the interview is over (the AI says "The mock interview is now complete"), close the voice mode screen. The entire conversation transcript will be saved in the chat. Paste the following grading prompt into the same chat to get your scorecard:

Now, analyze my answers from the mock interview. Evaluate each of my responses using the STAR (Situation, Task, Action, Result) framework.

Provide a table with three columns:
1. Question Asked: The question you asked.
2. Critique: A critical evaluation of my response. Focus on whether I clearly stated my action and provided quantifiable results (metrics).
3. Refined Answer: Rewrite my response in polished, senior-level language that maintains the same facts but sounds more impactful and clear. Ensure the tone is professional, outcome-oriented, and free of corporate jargon.

Grade my overall performance out of 10.

The Real World Story

A few months ago, a friend of mine in Bengaluru was preparing to switch companies and interviewing for a Senior Consultant role. He used ChatGPT's voice mode to practice speaking his answers aloud, but quickly realized he was wasting half his prep time listening to flattery. After every single response, the AI would gush: "Excellent answer! That shows great leadership. Wonderful job!" In a real partner round, no interviewer spends 60 seconds congratulating you after every sentence—the polite AI was creating a false sense of security.

To fix this, we set up ChatGPT using a strict "Skeptical Partner" persona, instructing it to remain in character and withhold all feedback until the mock interview was completely finished. In the next run, when my friend gave a vague answer about resolving a project delay, the AI didn't flatter him. It cut straight to the point: "That sounds like a team effort. What was your specific, single-point action to resolve the client conflict, and what was the final delivery impact in days?"

My friend is currently in the middle of active interview rounds. While he hasn't signed an offer letter yet, his situational delivery has improved dramatically. By silencing the AI's constant praise and forcing a strict step-by-step roleplay, he has trained himself to answer under pressure without rambling—meaning he walks into actual interviews prepared for the toughest questions they can throw at him.