How to Build an AI Co-Pilot to Ace Your PM Interviews | Ben Erez (ex-Meta)
Use Ben's proven product sense template to create your own AI interview co-pilot. Plus try his advanced version that feels like a "calibrated interviewer that's available 24/7"
Dear subscribers,
Today, I want to share a new episode with Ben Erez.
Ben is a former meta PM who has helped hundreds of people land great jobs. One of the most unique things that Ben has built is an AI co-pilot that his students say feels like having a "calibrated interviewer that's available 24/7." I got him to show me exactly how to build this AI co-pilot using his proven templates and prompts.
Watch now on YouTube, Apple, and Spotify.
Ben and I talked about:
(00:00) The current state of the PM job market
(02:07) The exact structure of the product sense interview
(08:58) How to build your AI interview co-pilot step-by-step
(12:36) Live demo: AI co-pilot walking through a Meta interview question
(13:33) Why Ben designed the AI to be the candidate, not the interviewer
(25:23) How to prioritize the right user segments and problems to solve
(32:13) The biggest mistakes candidates make when challenged by interviewers
This episode covers how to build a simple AI co-pilot based on Ben’s free product sense template. Check out Ben’s advanced interview AI co-pilot for help prepping Meta-style product sense and analytical thinking interviews.
🎙️ Coming up next on Behind the Craft
Cursor Tutorial for Beginners from Cursor’s Head of Education | Lee Robinson
AI Evaluations Explained: Step by Step with Real Examples | Aman Khan
How to Automate Your Work Using Claude Code | Alex Finn
All 100% free, subscribe if you like these practical, hands-on tutorials 🙂
Top 10 takeaways I learned from this episode
Find your candidate-market fit. Companies like Vercel get 3,000+ applications for each role. So instead of casting a wide net, target specific roles that match your background like "L7 AI trust and safety product lead."
Treat product sense interviews like a game with clear rules. Most candidates fail because they treat it like a casual conversation. Instead, follow a strict framework of motivation → segmentation → problems → solutions and manage your time.
Spend 70% of interview time on users and problems, not solutions. The solution phase should be around 10 minutes in a 35-minute interview. Interviewers want to see how you think through strategy and prioritization.
Brainstorm three options at each stage. Whether picking user segments, problems, or solutions, three is the magic number. "Four is too many, two doesn't show breadth" and interviewers want to see your creativity.
Build a simple AI interview copilot using Projects. Set up a ChatGPT or Claude project with some questions from Lewis Lin's question bank, Ben’s product sense post for Lenny, and his product sense template below with step by step guidance: