Everyone wants to build "AI-powered" apps, but few talk about the invoice that arrives at the end of the month.

With the massive hardware and energy costs required to run large foundational models, sustaining a cloud infrastructure for LLMs is often impossible if you are launching a side project or running an indie studio.
In this session, we shift the focus from the cloud to the device by exploring Small Language Models (SLMs).

We will explore the current state of local AI models for Android, walking through the integration of on-device functionality using real-world examples. We will compare the ease of using the built-in Android AICore with Gemini Nano against the flexibility and tool-calling capabilities of deploying models like Gemma 4 using Google AI Edge and LiteRT-LM.

Finally, we’ll take a look at what’s coming next as Android evolves into an "Intelligence System", and how the new AppFunctions will let you plug your app's features directly into on-device AI agents.