Android Developers: Agentic coding workflows with small models

While large frontier models dominate the headlines, small language models (SLMs) like Gemma 4 offer low-latency, cost-effective, and fully offline intelligence to augment daily developer workflows.



This talk explores practical patterns for integrating small models into Android development, from smart task routing to cost optimization, using Gemma 4 as our primary hands-on case study to demonstrate real-world use cases and how to configure developer tools and agentic workflows.



Speakers: Ian Ballantyne, Jose Alcérreca, Zoe Lopez-Latorre



Chapters:

0:00 - Introduction to small language models

8:42 - Multi-token prediction and speculative decoding

11:13 - Building custom evaluations

15:30 - Coding workflows with small language models

17:35 - Evaluating models with benchmarks

24:19 - Practical tips and tricks

27:25 - Multi-model orchestration

29:17 - Key takeaways and recommendations



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 2066



Products Mentioned: Gemma 4, Gemma for Android, Developer Tools