AI Meetup, Wroclaw with Callstack & Vercel

The AI SDK provides an elegant abstraction layer that simplifies building sophisticated AI applications across cloud and on-device environments. AI SDK allows providing custom providers - and in this talk, we will demonstrate how we built a custom provider and brought on-device local offline LLMs to mobile apps.

Join us as we explore the creation of react-native-ai, a custom AI SDK provider that enables local LLM execution directly in React Native apps. We'll dive deep into the provider architecture of AI SDK and show how we leveraged it to integrate with MLC LLM Engine and Apple's foundation models to mobile devices.

What you'll discover:

• AI SDK: why we decided to rely on it when building react-native-ai?
• AI SDK Provider Architecture: Understanding the core abstraction and how providers work
• Building Custom Providers: Step-by-step implementation of on-device LLM providers
• Seamless Integration: How ⁠react-native-ai plugs into ⁠generateText() and ⁠streamText()
• Apple Intelligence Provider: Creating ⁠@react-native-ai/apple as an AI SDK-compatible provider
• Provider Flexibility: Switching between cloud and on-device models with identical APIs
• Live demo: we will showcase how the same AI SDK code works seamlessly across OpenAI's cloud models and local on-device LLMs, highlighting the power of AI SDK. Learn how to extend AI SDK's capabilities to new platforms and unlock privacy-first, offline-capable AI applications.