The PM Playbook for the AI Era: Robby Stein on Taste, Decisions and Craft

Key points

Key takeaways from Lenny's Podcast (Lenny Summit) talk by Robby Stein, VP of Product Management at Google (September 2026):

PM craft is decision making. The real value of product management in the AI era is no longer getting things done but judgment and taste: being the absolute best at making decisions.

Understand people deeply first. Inspired by Clayton Christensen's Jobs to Be Done: discover the job a product is hired for through the user's story — the bed/mattress transference insight, and the unmet AI-era need like multimodal couch-shopping journeys.

Diagnose root cause, not symptom. Reframe failures: launch a version, ask why people aren't adopting, cluster leaf-node answers by theme, quantify weights with large surveys — then fix the biggest root cause, iterate recursively like training epochs.

Cut the product back to what works. Instagram Stories only won after cutting it out of the feed entirely; the Close Friends concept was confusing as a feed/favorites feature until relaunched as a Stories experience.

Reels flipped the format assumption. Ephemeral was wrong: creators wanted their content to live permanently, become a business and go viral — a misread only corrected with a large behavior-driven update.

Scale craft with AI. Self-built agents (on Google's anti-gravity system) QA the product at scale — evaluate search responses against a rubric, flag broken math/LaTeX, missing visual trays, and increasingly fix off-spec behaviour themselves.

Delight is the second half of craft. The search box's cycling gradient cursor, loading ring and haptics communicate humanity — the work that makes a product feel cared-for and gives it a soul.

The AI-era playbook is still the same craft. Understand people deeply, be rigorous about root causes, and build something awesome, simple and delightful — the same three chapters describe great products and great PMs alike.

Read more: YouTube