WWDC 2026 Developer Tools: Foundation Models, Core AI, Xcode

Apple's new developer tools include Foundation Models framework, Core AI for local model hosting, and agentic coding in Xcode. Build AI apps with on-device privacy.

By Subhendu Mahapatra · · 10 min read
WWDC26DevelopersXcodeFoundation ModelsCore AISwiftApple

WWDC 2026 wasn't just about consumer features. Apple quietly rolled out a comprehensive set of developer tools that make it easier than ever to build AI-powered apps. From Foundation Models to Core AI to agentic coding in Xcode, the developer story is just as compelling as the user story.

If you're building apps for Apple platforms, these tools change the game.

> [!PULL] > Apple's new Core AI framework lets you run other models locally in your app with the full power of Apple silicon. You're no longer limited to Apple's own models.

---

Foundation Models Framework

Apple's Foundation Models framework got significant upgrades this year. It's the primary way to access Apple's on-device AI capabilities in your apps.

Image Input

The biggest addition: you can now use images as input in addition to text. This opens up use cases that weren't possible before:

  • Analyze photos and provide insights
  • Generate descriptions of visual content
  • Process screenshots and documents
  • Understand charts and graphs
  • The example Apple showed was an app called Daydream that uses the on-device model to identify clothing items in a photo and provide fashion recommendations.

    Custom Skills

    You can now extend the Foundation Models with custom skills. These are specialized capabilities that add domain-specific knowledge to the base model.

    Want your app to understand medical terminology? Add a medical skill. Need it to process legal documents? Add a legal skill. The base model handles general tasks, while custom skills handle specialized ones.

    Server Models

    You can now use models running on servers through the same Swift API. This means you can start with on-device processing for speed and privacy, then fall back to more powerful server models when needed.

    The API is consistent whether you're running on-device or on-server, so you don't need to write different code for each scenario.

    ---

    Core AI Framework

    This is the announcement that might matter most for developers. Core AI is a new framework available on all Apple platforms that lets you run other models locally in your app.

    What This Means

    Previously, if you wanted to use a non-Apple model in your app, you had to either: 1. Send data to external servers (privacy concerns) 2. Bundle the model with your app (size concerns) 3. Use Apple's models only (limitation concerns)

    Core AI solves this by providing a standardized way to run models locally on Apple silicon. You can use models from other providers, run them on the user's device, and take advantage of the Neural Engine for acceleration.

    Why This Matters

    This is Apple's answer to developers who want to use specific models for specific tasks. Need a specialized coding model? Run it locally. Want to use a particular language model? Run it locally.

    The key advantage: privacy. Your users' data never leaves their device. This is a significant differentiator from cloud-based AI services.

    > [!PULL] > Core AI means you can bring your own models to Apple platforms while maintaining user privacy. The data stays on-device.

    ---

    Xcode: Agentic Coding

    Xcode is getting some major upgrades that change how you build apps.

    Agentic Coding Assistant

    The Xcode coding assistant can now:

  • Localize entire apps: Translate your app to multiple languages in one go
  • Interact with simulated devices: Test your app in realistic scenarios
  • Extend with custom skills: Add specialized capabilities
  • Choose your model: Pick the AI model that works best for your use case
  • Model Flexibility

    You're no longer limited to Apple's models. Xcode now supports Gemini and other models, giving you flexibility in how you build and test your apps.

    Tool Integration

    Xcode integrates with popular developer tools:

  • Figma: Import designs directly
  • GitHub: Manage repositories and pull requests
  • Custom tools: Connect your own development tools
  • Device Hub

    The all-new Device Hub brings every device — simulated and real — into one unified interface. You can:

  • Simulate multi-touch controls (swipe, pinch)
  • Change your app's appearance in one click
  • Resize dynamically to iterate like never before
  • This eliminates the need to switch between different simulator windows and physical devices.

    ---

    App Intents and Siri Integration

    Apple is pushing developers to adopt App Intents to integrate with Siri AI. The examples they showed demonstrate the potential:

  • Line (messaging app): Users can ask Siri to find information from conversations
  • Structured (calendar app): Users can ask Siri to create events and add them to their timeline
  • If your app indexes content into Spotlight, users can ask Siri to find information from it. If your app adopts App Intents, users can ask Siri to take actions in it.

    ---

    The Development Workflow

    Apple's vision for development is becoming clearer:

    1. Build with Foundation Models: Use Apple's on-device models for general tasks 2. Extend with Core AI: Add specialized models for specific use cases 3. Integrate with App Intents: Make your app accessible through Siri 4. Test with Device Hub: Verify across all device types 5. Ship with confidence: Leverage Apple's privacy framework

    This workflow is more integrated than ever, and the tools are designed to work together seamlessly.

    ---

    What This Means for the Ecosystem

    These tools signal Apple's commitment to becoming a serious AI platform. By providing:

  • On-device model execution (Foundation Models)
  • Local model hosting (Core AI)
  • Flexible development tools (Xcode)
  • System integration (App Intents)
  • Apple is creating an environment where developers can build sophisticated AI applications without sacrificing user privacy.

    This is a direct response to the cloud-centric approach of competitors. Apple is betting that developers will prefer on-device processing for privacy, performance, and offline capability.

    ---

    The Catch

    These tools require the latest Xcode and Apple silicon Macs. If you're developing on older hardware, you won't get the full benefits.

    Core AI is also new, so the ecosystem of available models is still limited. It will take time for developers to port their models to the framework.

    ---

    The Bottom Line

    WWDC 2026's developer tools are the most comprehensive AI development kit Apple has ever released. Foundation Models, Core AI, and Xcode's agentic coding create a powerful ecosystem for building AI-powered apps.

    The privacy angle is particularly compelling. Users get intelligent features without sacrificing their data. Developers get powerful tools without worrying about cloud costs or privacy regulations.

    If you're building for Apple platforms, these tools are worth learning now. The apps you build with them will define the next generation of iOS, macOS, and beyond.

    ---

    Related Reading:

  • The Complete WWDC 2026 Recap: Every Announcement Ranked
  • Building Agentic Loops on iOS 27: A Guide to the Siri Extensions API