Gemini product lead Tulsee Doshi on why product judgment matters more in the AI era
dotey · x · 2026-11-09
On Lenny's Podcast, Google frontier model product lead Tulsee Doshi argues AI lowers implementation cost but not the cost of picking the wrong direction — deciding what to build matters more than whether it can be built.
- Discovering new model capabilities: training produces unexpected capabilities; PMs shouldn't rely on leaderboards but use models hands-on to find standout tasks worth productizing
- Judgment via evals: start with 5-20 prompts you truly care about, judge results yourself, define what "good" means, then expand; retest prompts and patch rules after model upgrades — old workarounds may now be redundant or limiting
- 60/20/20 allocation: 60% on capabilities stable today, 20% betting on near-term model progress, 20% exploring emerging-but-immature experiences
Related event: Google Frontier Model PM on Product Judgment in the AI Era(2 posts)→
More from Companies & People
- OpenAI exec Lori not leaving, say industry watchers amid rumors — basedjensen · 2026-11-09
- Is Jensen Huang's CEO job at risk? Yale's Sonnenfeld estimates 'about a 50-50' — TechCrunch AI · 2026-11-09
- X user reports seeing ads despite Premium subscription, change unconfirmed — esrtweet · 2026-11-09
- NVIDIA AI Day Seoul Draws 2,000+ Developers to Showcase Physical AI, Robotics and Agentic AI — NVIDIAAI · 2026-11-09
- NVIDIA powers Korea's physical AI: Samsung sims 1600x faster, LG robots in factories — NVIDIA Blog · 2026-11-09
- NVIDIA and Korea's MSIT expand sovereign AI deal, opening NVLink Fusion to NPU makers — NVIDIA Blog · 2026-11-09