TL;DR
You just walked out of an interview. The next one is in 4 days. Open ChatGPT (GPT-5.5, free tier is enough) or Claude (Sonnet 4.6), paste the prompt below with your raw notes, and force a debrief that names the single biggest issue, two fixable patterns, and the one thing to drill — no “you did great.” Do it within 90 minutes while your memory is still sharp. The whole thing takes one prompt and one refinement pass.
When this is the right job for AI
- You took at least 10 lines of notes during or right after the interview.
- You can name 1-2 questions you fumbled.
- You want structured feedback, not therapy.
If you didn’t take notes, the next 4 days are better spent on prep than on reconstruction. AI can’t debrief a round you can’t remember.
What to feed the AI
A debrief is only as good as the input. Give it all six fields:
- Company + role + round number (“3 of 5, system design”)
- Topics covered (5-10 bullets, exact wording where you remember it)
- Questions you felt strong on
- Questions you fumbled, with what you actually said
- Any specific feedback the interviewer dropped
- Your read on the interviewer’s engagement (asked follow-ups? checked their phone?)
The engagement read matters more than people expect. A long answer that lost the room is a different problem from a long answer that held it.
Which model to use
Either of the two flagship chat models handles this well as of June 2026. Pick on workflow, not raw capability:
| Model | Best for this task | Cost |
|---|---|---|
| ChatGPT (GPT-5.5) | Voice mode to debrief out loud on your walk home; fast iteration | Free tier works; Plus $20/mo for higher limits |
| Claude (Sonnet 4.6) | A persistent Project that remembers your weak patterns across rounds | Free tier works; Pro $20/mo ($17 annual) |
If you are running a multi-round loop with the same company, Claude Projects is the better fit: it keeps your earlier debriefs in context, so by round 4 it can flag a pattern you have now repeated three times. For a one-off debrief, either is fine.
Copy-ready prompt
You are debriefing a job interview round to find the 2-3 things to fix before the next round.
Round: 3 of 5, system design, target role: Senior PM, Acme Cloud
Topics covered:
- Design a quota / billing system
- How to prioritize the first 3 features
- How to communicate trade-offs to engineering
- Behavioral: tell me about a time I disagreed with eng
Felt strong: trade-off communication, behavioral
Fumbled: prioritization (I jumped to features without framing user impact); the answer on "what would you ship first" went 4 min
Interviewer feedback dropped: "interesting framing on quota" + "let's come back to the prioritization question" (which we didn't)
Engagement: leaned in early, looked at notes during my long answer
Write:
1. The single biggest issue. One line. Direct.
2. Two fixable patterns (not "be more concise" — pattern level).
3. The exact thing to drill before round 4.
4. The story I should pre-build now in case it comes up again.
No "you did great" framing. Treat me like a peer who asked for honest feedback.Sample output structure
- Biggest issue: you bypassed framing on prioritization. The interviewer wanted to see your decision process; you gave them an answer.
- Patterns to fix:
- When the question is “how would you prioritize”, start with the framework you’re going to use (e.g. impact x effort x strategic fit). Do not skip to the conclusion.
- The 4-minute answer is a signal you didn’t check the interviewer’s eyes. After 90 seconds, pause and ask “want me to keep going or take a different angle?”
- Drill before round 4: practice 3 prioritization scenarios out loud with a 90-second cap and a forced framework step.
- Pre-build: “a time I changed my mind on a feature priority” — your prioritization answer needs a story to anchor the framework, structured as Situation / Task / Action / Result.
How to refine
One refinement pass usually gets you there:
- AI too soft → add: “this is honest feedback, not encouragement. Be willing to say
you missed a question.” - Output too generic → strict rule: “every fix must reference something specific the interviewer said or did.”
- Missing pattern level → add: “do not say
be more concise— name the conversational habit that produced the 4-minute answer.” - Want a pressure-test → add: “list the two follow-up questions I’m most likely to get wrong next round, based on these answers.”
Common mistakes
- Debriefing only the things that went wrong. The strong moments tell you what to repeat.
- No timeline. “Fix this eventually” doesn’t survive the next 4 days.
- Skipping the interviewer-signal field. Engagement cues are half the data.
- Letting AI list 8 things. Two patterns is the realistic upper bound for what you can actually fix in 4 days.
- Practicing the fix only in your head. Text prep has a ceiling; you have to say the new version out loud before it sticks (this is where ChatGPT voice mode earns its keep).
FAQ
- How long after the interview should I do this? Within 90 minutes — before your memory smooths the rough edges out and you start remembering the version you wish you’d said.
- Should I share the debrief with the recruiter? No. The debrief is for you. The recruiter gets a thank-you note, not a self-assessment.
- Is the free tier enough? Yes. A single debrief is one prompt plus a refinement pass — well inside the free limits on both ChatGPT and Claude as of June 2026. Paid tiers ($20/mo) only matter if you run many rounds in a day or want voice mode unmetered.
- Will an AI debrief leak my interview to my employer? No. Pasting your own notes into a chat does not contact the company. Standard caution applies: don’t paste anything under NDA, and turn off chat training in settings if the content is sensitive.
- Can I use this for the final-round debrief too? Yes — same template, swap “fix before next round” with “pre-decision reflection: would I take this role, and what do I still need to learn about the team?”
For the structure your fixes should follow, the STAR method (Situation, Task, Action, Result) is the standard backbone for behavioral answers and a good target format when you pre-build a story.
Related
- AI behavioral interview prep
- AI mock interview
- AI tell-me-about-yourself practice
- AI thank-you email after interview
- Interview Debrief Prompts for Lessons From Every Round
Tags: #AI writing #Job search #Interview #Interview prep #Story bank