Take your Android AI features to the next level by building autonomous on-device agents. In this session, we dive into ML Kit’s new agentic APIs, built directly on top of the direct model access APIs you already know.
Using the open-source Jetpacker app, you’ll learn how to chain multi-step actions across local data, manage stateful multi-turn sessions with isolated contexts, and wire up Android MCP to power both in-app assistants and Android system-level surfaces with a single codebase.
Resources:
Android ML Kit overview → https://goo.gle/4kd3hM1
Jetpacker samples → https://goo.gle/4iBaynK
Speakers: Jolanda Verhoef, Victoria Lestari
Chapters:
0:00 - Introduction to on-device AI and Gemini nano
5:46 - Jetpack agent demo and function calling
10:49 - Building and implementing an AI agent runner
14:06 - Multi-step agent architecture and execution
20:01 - Managing multi-turn conversation sessions
24:17 - Advanced agentic workflows and tool integration
30:50 - Best practices and conclusion
Watch the intelligent apps sessions from Android Dev Summit 2026 → https://goo.gle/ADS26-Intelligent-Apps
Watch all the sessions from Android Dev Summit 2026 → https://goo.gle/ADS26-All-Sessions
Subscribe to Android Developers → https://goo.gle/AndroidDevs
#AndroidDevSummit2026
Event: Android Dev Summit 2026
Products Mentioned: ML Kit, ML Kit GenAI, Android MCP