In the latest episode of the podcast Google AI: Release Notes, Logan Kilpatrick talks with Emanuel Taropa about how a team called Smokejumpers makes it possible for Gemmini 3 to reach billions of people. Can you imagine the coordination behind an AI feature that millions use at the same time? Here I explain why it matters.
What the Smokejumpers team is
Smokejumpers is an agile, cross-functional team inside Google that handles the most complex and critical AI launches. It's not just a group of engineers: it brings together product, operations, infrastructure, and high-risk technical bets.
In the January 27, 2026 conversation, Emanuel Taropa and Logan describe how they work together to design, test, and deploy models like Gemini 3 without interrupting services used by billions.
How they bring Gemini to scale
Three pieces stand out in the episode and make the difference:
-
The technical connectivity between services. A great model isn't enough; you need a layer that makes it reliable, fast, and observable in production. Think of a set of pipes that connect data, inference, and monitoring.
