Discover a new app development paradigm where users interact naturally with apps, blurring the lines between human language and digital action. We’ll demo newly released libraries, which allow local agents to drive applications and devices by combining on-device LLMs with native audio input, retrieval augmented generation, and function calling. Learn how to accelerate on CPU, GPU, or NPU, and a tutorial on how to test ML workflows on real devices via the cloud using Google AI Edge Portal. The session will include real production features from our current 3P EAP partners. Resources: For more information → https://goo.gle/ai-edge LiteRT acceleration updates → https://goo.gle/3FnDGOr GPU Updates → https://goo.gle/3ScpzhS NPU EAP Signup → https://goo.gle/44NuNbo Sample App → https://goo.gle/4mqk9O9 AI Edge Portal → https://goo.gle/43F58k8 AI Edge Portal EAP Signup → https://goo.gle/436hYaT Speakers: Matthias Grundmann, Mark Sherwood, Mogan Shieh Check out the AI session track from Google I/O 2025 → https://goo.gle/io25-ai-yt Check out all of the sessions from Google I/O 2025→ https://goo.gle/io25-sessions-yt Subscribe to Google for Developers → https://goo.gle/developers Event: Google I/O 2025 Products Mentioned: AI/Machine Learning