Apple Rebrands Siri as ‘Siri AI’ at WWDC 2026, Turns Every App Into an Agentic Endpoint
Apple rebranded its voice assistant as “Siri AI” at WWDC 2026 on Monday, shipping a standalone conversational app powered by Google Gemini that can chain multi-step actions across apps, retain context across conversations, and act autonomously on behalf of users. The overhaul, first promised in 2024, represents Apple’s formal entry into the agentic AI race.
What Shipped
The core product is a dedicated Siri app, synced across iPhone, iPad, Mac, Apple Watch, and Vision Pro via iCloud, according to CNBC. Users can revisit old conversations, ask follow-up questions with full context retention, and get longer, more detailed responses. “Siri is now a profoundly more capable assistant,” Apple VP Mike Rockwell said during the keynote. “It’s also more conversational, so you can go back and forth like never before.”
Apple also launched App Intents 2.0, a framework that lets any third-party app register as an endpoint Siri can call without the user opening it. As FourWeekMBA noted, this effectively turns every compliant app into an agentic tool: Siri can book a restaurant via OpenTable, check traffic in Maps, and text someone the ETA in a single chained command. Apple demonstrated the pattern with its own Passwords app, which now goes to individual websites to change insecure credentials autonomously.
The Gemini Partnership
Google Gemini provides the underlying model, the result of a multiyear deal announced in January. Apple’s most advanced cloud model, AFM Cloud Pro, runs on Nvidia GPUs in Google’s infrastructure, according to CNBC. Craig Federighi, Apple’s SVP of Software Engineering, framed the integration around on-device processing: “Privacy in AI is nonnegotiable,” he said at the keynote, per PYMNTS.
iOS 27 will also open Siri to third-party extensions from ChatGPT and Anthropic, according to Tom’s Guide, giving users a choice of AI providers within a single native interface. Apple is building the orchestration layer through which every model gets accessed — the control plane sitting between users and whichever model they prefer.
Availability and Constraints
Siri AI launches in English beta later this year in the U.S. It will not be available in the EU or China at launch due to regulatory requirements, according to Variety. Hardware requirements are steep: iPhone Air, iPhone 17 Pro/Pro Max, iPad (M4+) or Mac (M3+) with at least 12GB unified memory.
The Agent Orchestration Angle
Apple’s play is not about building the best AI model. It is about controlling what sits between the model and the user. PYMNTS Intelligence data found that Android users are 24 percentage points more likely to use Google Gemini than iOS users, because Gemini ships as the default on Android. ChatGPT, meanwhile, is the preferred AI on iOS by 10 points over Android. Apple users were bypassing Siri to reach the experience they wanted.
App Intents 2.0 changes that calculus. If every app on iOS can be called as an agentic endpoint through Siri, developers who don’t implement the framework become invisible in what FourWeekMBA calls “the agentic App Store.” The parallel to Microsoft’s Build 2026 announcements last week, where MXC containers and Agent Governance created an OS-level agent infrastructure on Windows, is direct. Both companies are making the operating system the control plane for AI agents, not the model providers.
Apple reported record Q2 2026 services revenue of $31 billion, up 16% year over year, per PYMNTS. The App Store takes 30% of AI subscriptions in year one, 15% thereafter. If Siri becomes the primary way users access AI on iOS, Apple collects a toll on every agent interaction without training, hosting, or supporting a model.