All titles verified real (author/year/publisher), July 2026. Reading order matters less than matching the book to your current gap.
| # | Book | Author (Year) | The job it does |
|---|---|---|---|
| 1 | The AI Product Manager's Handbook (2nd ed.) | Irene Bratsis (Packt, 2024) | The most direct "how to do this job" playbook, updated for genAI + responsible AI |
| 2 | AI Engineering: Building Applications with Foundation Models | Chip Huyen (O'Reilly, 2025) | How modern LLM products are actually built: evals, RAG, agents; the most-read book on O'Reilly in 2025. Read it to speak eng fluently; skim the code |
| 3 | Co-Intelligence: Living and Working with AI | Ethan Mollick (Portfolio, 2024) | Mental models for working with LLMs; the book to hand your exec |
| 4 | Prediction Machines | Agrawal, Gans & Goldfarb (HBR Press, 2018) | The economics: AI = cheap prediction; where cheap prediction changes decisions. Aging well |
| 5 | Inspired (2nd ed.) | Marty Cagan (Wiley/SVPG, 2017) | The PM craft canon: discovery, empowered teams; AI PM is PM first |
| 6 | Cracking the PM Interview | McDowell & Bavaro (2013) | Still the broadest interview-prep base; layer our AI question bank on top |
"I need deeper ML/product systems thinking" → Designing Machine Learning Systems, Chip Huyen (O'Reilly, 2022): the production-ML standard · Building Machine Learning Powered Applications, Emmanuel Ameisen (O'Reilly, 2020): idea→shipped ML product, explicitly PM-friendly.
"I need the math/architecture intuition, gently" → The StatQuest Illustrated Guide to Neural Networks and AI, Josh Starmer (2025): the gentlest real explanation of how neural nets/LLMs work (pairs with his YouTube).
"I'm interviewing" → Decode and Conquer (5th ed., May 2025), Lewis C. Lin: 528 pages; the new edition adds AI-fluency and technical-judgment sections · Cracking the PM Career, McDowell & Bavaro (2021): leveling and growth · Swipe to Unlock, Mehta/Agashe/Detroja: tech concepts for non-engineers, fast interview prep for tech-context questions.
"I lead teams / set strategy" → Empowered, Cagan & Jones (Wiley/SVPG, 2020): the leadership companion to Inspired · plus re-read Prediction Machines with your portfolio in mind.
- Pair books with builds: read AI Engineering's eval chapter, then do Project 4; concepts anneal when applied within 48 hours.
- Books age in this field asymmetrically: economics (Prediction Machines) and craft (Inspired) hold for years; tooling chapters stale in months. Treat any specific model/vendor mention as a snapshot and cross-check with current tools.
- The newsletter layer is where the field actually moves: books for foundations, newsletters for the live edge.