Key Insights: 'How to Vibe Code in PM Interviews'
+ Session Recording
The 3rd cohort of Advanced Tech & AI Program is starting 20th September. We're building production-ready AI products in 8 weeks, along with learning advanced AI theoretical topics. You can learn more and apply to the program here
Just wrapped up a live session on "How to Vibe-Code in PM Interviews with AI" last Saturday and wanted to share the key insights.
For context - this is the first episode of my "How to X with AI" series. You can follow updates by subscribing to this newsletter.
Here are the key insights from the session:
Certain aspects of PM role are changing fast
Google's recent shift from "writing-first" to "building-first" culture isn't just a Google thing. It's happening everywhere.
Vibe-Coding is now a core PM skill
Think of it like user segmentation or success metrics - it's becoming table stakes. Companies aren't testing if you can use tools like Claude or Lovable. They're testing if you can think like a PM while building these prototypes.
The Framework for Vibe-coding
Jumping straight into prototyping is a mistake, both in full time role and in interviews. Here's the framework we discussed in the session, it’s pretty similar to product sense/ design problems:
Most PMs miss this. They go straight to building the prototype ("creating a feedback form") without asking WHO it's for or WHY we're building it.
Being Tool Agnostic
You need 2-3 diverse tools in your arsenal. For example,
Lovable for building end-to-end apps
Claude for reasoning and creating clean UI
Replit for AI agent development of apps
But I would ask you to remember that you shouldn’t get caught up in things like prod vs testing environment, or specific features. That’s is what the interviewer is testing you around. That’s easy stuff. They care if you can show product sense, build quick prototypes, focus on what matters, and debug when things go wrong.
Most Prompt Engineering advice is shallow for PMs
Most prompt engineering advice comes in the form of "give it a persona" - “be clear and specific” - “add examples”. PMs have to get complex things done. The advice here is to go deeper. Read the actual research papers on techniques such as Chain of Thought, Tree of Thought, etc.
Show Thoughtful PM Skills, Not Speed
The best signal you can give to the hiring manager isn’t how many features you know of a tool. Comment on your design choices while building. Fact-check LLM outputs. Show how you handle when AI goes haywire.
When I showed how to solve the design problem “Build a feedback form that dynamically adapts to user feedback”, I spent time critiquing the LLM's output - "too many response options", "users won't understand the term 'order fulfillment” etc. That's what separates good PMs from prompt copy-pasters.
The session recording is available here
Summary
Vibe coding interviews are about testing if you can 3-4x your productivity as a PM. Not if you can build pretty UIs.
Master the fundamentals - product sense, technical system design, RCA skills. Then layer AI tools on top. Not the other way around.
Next session in the series drops next week. You can register here: Next session in the series drops next week. You can register here: https://maven.com/p/af7cd0/how-to-vibe-code-in-pm-interviews-part-2-of-2
You can learn the AI prototyping end-to-end by attending this free course: https://www.pmcurve.com/program/prototyping-with-ai-for-pms-founders
---
If you want to go beyond how to use AI as a PM, and build actual AI products as a product leader, do check the advanced AI program: https://www.pmcurve.com/program/advanced-ai-for-pms
Thanks,
Deepak