Google's AI Strategy: Full-Stack Magic for Diverse Use Cases

Nov 8, 2026 · Lenny's Podcast
🎧 PodShort 58 min squeezed to 2 AI SprinklerAS Sales Tech New
Episode artwork
Tulsee Levins
Head of Product for Frontier AI Models at Google
Lenny's Podcast
58 min squeezed to 2
Full episode from Lenny's Podcast
Quotable Moments

The model is not predictable. It's kind of like Snape's potions class. You can put all the ingredients in, but you also have to expect that there will be a little bit of magic.

Tulsee LevinsHead of Product for Frontier AI Models at Google
00:01

You want to actually build features that aren't necessarily always working today. It feels like every AI assistant launching these days is taking all this juicy Google data, my Gmail, Google Calendar, my Docs, my... I'm curious just why Google hasn't launched something awesome yet.

Tulsee LevinsHead of Product for Frontier AI Models at Google
00:47

The model is not predictable. You can put all the ingredients in, but you also have to expect that there will be a little bit of magic. And that magic is going to be surprising.

Tulsee LevinsHead of Product for Frontier AI Models at Google
03:16
Key Insights
  • AI model development is not entirely predictable, often involving 'magic' and surprising outcomes, similar to a complex potion-making class.
  • When hiring PMs for AI products, it's crucial to look for individuals who understand model dynamics, have strong opinions, and are comfortable experimenting with models daily.
  • Building AI models involves not just defining desired outcomes but also embracing the unpredictable nature of the technology and identifying where the model uniquely shines.
  • Google's AI strategy is full-stack, from hardware to models to end-user applications, aiming to serve a diverse range of user needs across different scales and use cases.
  • The 60/20/20 rule, which emphasizes building a core functional product, future-proofing for immediate needs, and experimenting with novel capabilities, guides AI product development.
  • Safety in AI is a continuous process, not a one-time fix, and requires robust monitoring, intervention, and a proactive approach to patching vulnerabilities.
  • The distinction between AI 'growth' and 'building' is crucial; models grow organically, but their application and refinement require intentional design and human oversight.
  • Google's approach to AI is to build a breadth of experiences, not just focus on the most advanced models, catering to diverse users from everyday consumers to enterprise engineers and defenders.
Metrics Mentioned
  • 14 products above a billion users (Google's scale and reach across its product portfolio, underscoring its full-stack AI strategy.)

RevBots.ai View:

  • Google's approach exemplifies the AI Sprinkler stage: AI features bolted onto existing infrastructure.
  • Full-stack AI strategy highlights the need for integration across hardware, models, and applications.
  • The 60/20/20 rule underscores the importance of balancing core functionality with innovation.
  • Continuous AI safety monitoring aligns with the need for proactive risk management in AI-driven GTM.
🎧Full Episode:Lenny's Podcast →