How do I implement foundation models in my app?

To implement Foundation Models in your app, the basic flow is:

1. Import the framework
2. Create a language model session
3. Send a prompt to the model
4. Optionally add:
   • Structured output with Generable
   • Streaming responses
   • Tool calling for app actions
   • Stateful sessions for multi-turn interactions

Apple showed that the simplest version is basically “three lines of code”: import the framework, create a session, and call respond with your prompt.

What you can build with it

• Text generation
• Summarization
• Text extraction and analysis
• Structured outputs
• Tool-using agents
• Personalized app experiences

Important notes

• It runs on device in Apple’s framework, so it’s private and can work offline.
• The framework is built into the OS, so it doesn’t increase app size.
• You can also use newer capabilities like guided generation and tool calling to make the model more useful in your app.

Good WWDC sessions to start with

• Meet the Foundation Models framework
• Deep dive into the Foundation Models framework
• Code-along: Bring on-device AI to your app using the Foundation Models framework

If you want, I can also give you a minimal Swift example of using a Foundation Models session in your app.