AI features are now expected in modern mobile apps, but building them well requires more than simply calling an API.

In this hands-on codelab, attendees will build an AI-powered chat experience on Android while learning how to integrate cloud models, run on-device inference, and connect everything into a real app architecture. By the end, participants will have a working feature and a clear understanding of how to apply these techniques in their own apps.

What attendees will learn:

- Integrate cloud-based AI APIs alongside on-device models within a single Android application
- Design effective prompts and manage multi-turn conversation context
-Implement streaming responses for real-time interactions and handle structured outputs
- Run local models using Android-compatible runtimes, including model download and update strategies
- Build a hybrid architecture that intelligently switches between cloud and on-device inference
- Optimize for real-world constraints such as latency, memory, compute, cost, and network availability
-Design resilient user experiences with proper handling of loading states, errors, retries, and cancellations
- Enable offline functionality and privacy-preserving inference

By the end of the codelab, attendees will have a fully functional AI chat feature and a clear, reusable architecture for building scalable, reliable, and modern AI-powered Android applications.