Apple's Siri Architecture Supports Third-Party AI Model Integration, Code Reveals

Third-Party AI Integration in Siri

Analysis of private frameworks in iOS 27 and macOS Golden Gate reveals that Apple's new Siri architecture is designed to integrate third-party AI models. This integration appears to be deep, allowing external AI services to function within the Siri ecosystem.

Model Delegation for Extensions

One mechanism, called Model Delegation, allows AI models like Claude to appear as Siri extensions. This functions similarly to the existing built-in ChatGPT extension. Users can select Claude as the AI model through a contextual menu in the "Search or Ask" bar. When a request requires an Apple system feature, Claude can hand the task back to Siri for execution, such as setting a reminder in the Reminders app.

Full Model Replacement via Inference Provider

A second protocol, the Inference Provider in "Model Manager Services," allows Apple's own server-side Siri model to be completely replaced by another model, such as GPT-5.6 Terra. In this scenario, the third-party model receives Apple's Siri planner prompt and tool definitions, enabling it to request system actions, access personal data, and formulate responses that Siri then presents using its own interface and voice. This was demonstrated with ChatGPT finding emails, summarizing them, and sending a message via the Messages app.

Implications for AI Landscape

This level of integration could significantly open up the AI landscape for Apple software requests, providing users with more advanced capabilities beyond Siri's native functions, such as creating CSV files. The European Union's Digital Markets Act, which mandates third-party access to iOS features, may have influenced Apple's development of this approach.