The Model Eats the Scaffolding: DeepMind's Logan Kilpatrick & Tulsee Doshi on 3.5 Flash, Omni & More
Google I/O launchesGemini 3.5 FlashAI agent infrastructureMultimodal AICost-adjusted efficiencyModel scaling strategyAgent harness developmentProduct integrationRecursive self-improvementAI research collaborationAudio input modalitiesContext window limitsDiffusion modelsModel psychologySearch grounding
This episode features Logan Kilpatrick and Tulsee Doshi of Google DeepMind discussing Google's AI strategy and new launches at Google I/O, including Gemini 3.5 Flash, the Omni video generation model, and the Gemini Spark agentic product. They delve into Google's focus on cost-adjusted efficiency, the development of a robust agent harness for product integration, and their perspective on recursive self-improvement and AI collaboration.