Google AI: Release Notes Gemini 3: Launch day reactions
16 snips
Nov 25, 2025 
Tulsi Doshi and Josh Woodward join to discuss the exciting launch of Gemini 3. Tulsi, a product lead for generative AI models, shares innovative capabilities like multimodal understanding and agentic features. Josh highlights Gemini's integration across Google surfaces for developer access. They dive into real-world applications, such as transforming handwritten recipes into interactive apps and rapid game development. The duo also reflects on balancing model performance with accessibility, and how user feedback drives continuous improvements.
AI Snips
Chapters
Transcript
Episode notes
Step Change In Capabilities
- Gemini 3 is a step change in reasoning, multimodal understanding, coding, and agentic use.
- The team positions it to “bring anything to life” across modalities and products.
Ship Widely To Learn Fast
- Ship models widely on day one to get fast feedback across real product surfaces.
- Use that feedback to iterate model behavior and prioritize product trade-offs.
Product Feedback Shapes Model Trade-Offs
- Tight product-model feedback loops surface concrete trade-offs between capabilities and constraints.
- Those trade-offs guide which revs are acceptable for different product contexts.