WWDC 2026: Apple Unveils Its New Gemini-Powered AI Architecture That Will Enable Siri AI

Apple kicked off its annual developer conference on Monday. This year's Worldwide Developers Conference (WWDC) started with the keynote presentation from the Cupertino-based tech giant, during which Apple unveiled its new iOS 27, iPadOS 27, macOS 27, visionOS 27, and watchOS 27. However, the keynote presentation also marked the reintroduction of the tech giant's new AI-powered voice assistance, Siri AI, which was first demonstrated during WWDC 2024. Earlier this year, the Tim Cook-led company partnered with Google, allowing Gemini to power the revamped Siri and Apple's foundational AI models. Now, the company has unveiled its new Gemini-powered AI architecture.

New Apple Intelligence Architecture Revealed

During the keynote presentation at the commencement of WWDC 2026, the iPhone-maker unveiled its new Apple Intelligence architecture, with AI models built on Google Gemini. Demonstrated with the help of concentric circles, Apple shows that its foundational models are situated at the core of its AI architecture, bringing new AI tools and features to iPhone, iPad, Mac, Apple Watch, and Apple Vision Pro models.

Apple's AI architecture is centred around its foundational AI models
Photo Credit: YouTube/ Apple

Built in collaboration with Google, Apple's foundational AI models support multimodality. This means that they are capable of generating responses based on text, voice, and image-based prompts. Moreover, users can take the help of these tools to generate and edit images using natural language prompts and other references.

Moving outwards, Apple's foundational models are also capable of generating responses based on personal context, further tailoring results best-suited to a user's preferences and past conversations. These AI models are also capable of extracting context from the screen of the device and generating responses based on what is being displayed. On top of this, the AI models are trained on publicly available data, bringing “World Knowledge” to them.

Similar to Gemini's multi-step task automation functionality on Android devices, Apple's foundational models are capable of taking actions on behalf of the users. For example, users can ask Siri AI about the upcoming FIFA World Cup and then ask the voice assistant-tuned-AI agent to look for tickets for a specific match. It is also capable of setting reminders when asked by the user.

Another core component of Apple's revamped AI architecture is a new system orchestrator, which allows Apple's AI tools and security features to work in synchronisation across Apple's native apps on its devices. This system orchestrator also allows the company's foundational AI models to provide responses based on the app being used or the task running in the background. Dubbed “System Experiences”, it also allows Siri AI and apps to work together.