Siri AI: Everything Apple Announced at WWDC 2026 — Features, Release, Compatibility

If you heard "Siri AI" during the WWDC 2026 keynote and asked yourself: "Is this just a rebrand, or is it actually something different?" — that question was the point. Apple's voice assistant had been called Siri for fifteen years — first as a standalone app in 2010, then as the defining voice interface of the iPhone. Changing the name after a decade and a half signals that this is not a minor update.

The short answer: it is both. Apple did not just change the name. It rebuilt the underlying architecture around large language models, onscreen awareness, personal context, and deep app integration via the App Intents framework. The Siri you knew — the one that set timers and looked up weather — still exists underneath. But what it can do now is categorically different.

If you are trying to understand what was announced, when you can use it, whether your device supports it, and what it means for developers, this article covers everything from the WWDC 2026 announcements and the latest Apple developer documentation.

Full disclosure: this article is based on Apple's WWDC 2026 keynote, Apple Developer documentation, the Apple Intelligence preview pages, and the developer beta materials available as of June 9, 2026. Some features remain marked as "future software update" in current developer docs and were not independently testable at the time of writing.

What Is Siri AI?

Siri AI is Apple's rebranded, LLM-powered intelligent assistant for iPhone, iPad, Mac, Apple Watch, and Apple Vision Pro. It replaces the previous Siri experience and represents the most significant architectural change to the assistant since its launch in 2011.

The core difference is how the assistant processes requests. Previous versions of Siri relied on a fixed intent-matching system: the user spoke a command, Siri classified it into a predefined intent slot, and a handler executed that intent. If the phrasing did not match the expected pattern, the request failed or fell back to web search.

Siri AI replaces that pipeline with a large language model-based reasoning layer. The assistant now interprets the meaning of a request rather than matching it against a template. That changes everything — from how naturally you can speak to it, to what kinds of multi-step tasks it can handle.

What did not change: Siri AI remains on-device for most processing. Apple continues to use its own foundation models rather than third-party LLMs like GPT or Gemini for the core assistant experience.

Key Features Announced at WWDC 2026

Apple highlighted four major capability areas for Siri AI during the keynote. Not all were available in the initial developer beta.

Conversational Experience

The most immediately visible change is natural, multi-turn conversation. You do not need to say "Hey Siri" before every follow-up — the assistant maintains context across a session, handles interruptions, corrections, and mid-sentence changes of direction.

The difference in practice:

During the keynote, Apple demonstrated a user asking about dinner reservations, then immediately asking "what about the one on Valencia Street?" without re-specifying that they meant restaurants, the city, or the time — Siri AI carried all of that context across turns.

Personal Context Understanding

Siri AI can access and reason across your personal data — messages, emails, calendar, photos, and files — to answer questions that previously required manual searching.

Examples from the WWDC demo included:

• "Find the email Alex sent last week with the venue contract attached" — surfaced an email from six days ago with the correct attachment
• "What time is my appointment on Thursday, and will traffic be bad?" — combined calendar data with real-time traffic routing
• "Show me photos from our trip to Portland" — searched the photo library by location and time, not just album names

Apple stressed that personal context processing happens entirely on-device. The data does not leave the device for processing, consistent with Apple's privacy architecture.

Onscreen Awareness

Siri AI can see and understand what is on the screen. If a user receives a text message with an address and says "add this to my calendar," Siri AI recognizes the address on screen, extracts it, and creates the calendar event — without the user needing to manually copy or re-type the information.

This extends to images, PDFs, web pages, and app screens. The assistant identifies text, locations, dates, phone numbers, and contact information displayed on the screen and can act on them.

System-Wide and App Actions

Rather than only controlling built-in Apple apps, Siri AI can trigger actions inside third-party apps that have adopted the App Intents framework. Apple's demo showed a user saying "book a ride to the airport for 6 AM tomorrow" — Siri AI opened a ride-sharing app, set the pickup time to 6 AM, and confirmed the destination, all within the assistant interface.

The full list of controllable action categories in the initial beta includes:

• Communication: Send messages, start calls, compose emails via any supported app
• Productivity: Create documents, add calendar events, set reminders
• Media: Play music, start podcasts, begin workouts
• Navigation: Get directions, book rides, check transit schedules
• Shopping: Reorder items, check order status
• Smart Home: Control HomeKit and Matter-compatible devices
• Health: Log symptoms, view trends, start workouts

Siri AI vs Apple Intelligence vs Original Siri

One of the most common points of confusion after the keynote is how these names relate to each other.

Apple Intelligence is the platform-level AI capability — the on-device models, the Private Cloud Compute infrastructure, the neural engine optimizations. Siri AI is one application that runs on top of that platform, alongside other Apple Intelligence features like Writing Tools, Genmoji, and Image Playground.

So Siri AI is not replacing Apple Intelligence. It is replacing the old Siri within the Apple Intelligence framework. If Apple Intelligence did not exist, Siri AI would not be possible — it depends on the same foundation models, the same on-device processing pipeline, and the same privacy architecture.

Does Siri AI Use Gemini or Third-Party Models?

No — not for the core assistant functionality. Apple has confirmed that Siri AI runs on Apple's own foundation models, processed either on-device or via Private Cloud Compute for requests that require additional compute power.

This has been a point of speculation since Apple and Google were reported to be in discussions about a Gemini licensing deal in 2024. As of the WWDC 2026 announcements, no such integration has been confirmed for Siri AI. Apple positions its own models — combined with the on-device processing advantage — as a privacy differentiator against competitors whose AI assistants depend on cloud-based third-party models.

However, Apple Intelligence already integrates ChatGPT for certain writing and image generation tasks in iOS 18/iOS 19. The key distinction: those integrations are opt-in, user-visible, and cover specific creative tasks. Siri AI's core reasoning — understanding what the user asked, deciding what action to take, retrieving personal context — stays within Apple's models.

Release Timeline and Availability

Siri AI was announced at WWDC 2026 on June 8, 2026. The developer beta became available the same day for registered developers. A public beta is expected in July 2026.

The general release is expected alongside this fall's major OS updates — tentatively called iOS 27, iPadOS 27, and macOS 27. Apple has not confirmed the exact naming of this year's OS releases, but the September timeline aligns with previous release patterns.

Important caveat: Some features demonstrated at WWDC are marked as "in development" or "future software update" in the current developer documentation. Apple developer Siri documentation notes that personal context understanding, onscreen awareness, and in-app actions are "in development" and may not ship in the initial public release. Apple appears to be staging the rollout: conversational improvements first, personal context and app actions later in the iOS 27 lifecycle.

Compatible Devices

Siri AI requires a device with a neural engine capable of running Apple's on-device foundation models. Based on the current beta and Apple Intelligence hardware requirements, the following devices are expected to be compatible:

iPhone:

• iPhone 17 Pro and iPhone 17 Pro Max
• iPhone 17 (A18 chip or later)
• iPhone 16 Pro and iPhone 16 Pro Max
• iPhone 16 (A17 Pro or later)

Note: iPhone 15 series (A16 Bionic) likely excluded — Apple Intelligence already required A17 Pro or M1 for full features, and Siri AI adds additional compute demands.

iPad:

• iPad Pro (M1 and later)
• iPad Air (M1 and later)
• iPad mini (A17 Pro)

Mac:

• MacBook Air (M1 and later)
• MacBook Pro (M1 and later)
• iMac (M1 and later)
• Mac mini (M1 and later)
• Mac Studio (M1 Max and later)
• Mac Pro (M2 Ultra and later)

Other:

• Apple Watch Series 10 or later (Apple Watch models with S9 or later chips for on-device processing; some Siri AI features will relay to paired iPhone)
• Apple Vision Pro

The device list mirrors the existing Apple Intelligence hardware requirements, with the addition of the latest 2026 product lineup.

Language and Region Support

Siri AI launches with the same language set that Apple Intelligence currently supports. Apple has consistently expanded language availability over the past two years, and Siri AI will inherit that footprint.

Some features — particularly onscreen awareness, personal context, and third-party app actions — may launch with narrower language support and expand over time. Apple's pattern with Apple Intelligence has been to ship new feature languages several months after the initial release.

China region devices will receive a separate version, as with previous Siri and Apple Intelligence releases, to comply with local regulations.

Privacy Architecture

Apple has presented privacy as the defining differentiator of Siri AI, and the architecture reflects consistent design choices across three layers.

On-device processing. The foundation models that power Siri AI run entirely on the device for the vast majority of requests. Personal context data — messages, emails, photos, calendar events — is processed locally and never sent to Apple servers. This includes all personal context queries, most conversational interactions, and all onscreen awareness processing.

Rule of thumb: If a request involves your personal data — messages, calendar entries, photos — it stays on your device. If it involves broad knowledge or complex multi-step reasoning, it may use Private Cloud Compute. The dividing line is simple: private data stays local; public reasoning can scale to the cloud.

Private Cloud Compute. When a request requires more compute power than the device can provide (complex multi-step reasoning, large-scale data retrieval), the processing is offloaded to Apple's Private Cloud Compute infrastructure. Apple has published the server software for inspection, and the architecture is designed so that data is used only to fulfill the request and is never logged or stored.

Opt-in for third-party models. Any integration with third-party AI models (ChatGPT for creative tasks, for example) requires explicit user opt-in. The user sees a prompt before any data leaves Apple's infrastructure. This is unchanged from the existing Apple Intelligence approach.

Apple has stated that Siri AI interactions are not used to train the foundation models.

The privacy architecture addresses what Siri AI protects, but the second major shift — one that determines how capable the assistant actually becomes over the next year — is how developers build for it. The transition from SiriKit to App Intents changes both what apps can expose to the assistant and, in turn, what users can ask Siri AI to do on their behalf.

Impact on Developers: App Intents Integration

For developers, the most significant change in Siri AI is the expansion of the App Intents framework. Apple has deprecated many of the older SiriKit intent structures and is consolidating around App Intents as the single way to expose app functionality to Siri AI.

What changed: Previously, SiriKit used a separate set of intent definition files that only exposed actions to the Siri interface. App Intents unifies the system: the same intent schemas work with Siri AI, Shortcuts, Watch complications, Spotlight, and Control Center. A developer writes the intent once and it surfaces everywhere.

What developers need to do:

1. Define App Intent schemas for actions the app supports
2. Provide natural language phrases that trigger each intent
3. Handle parameter resolution (entities, enums, relationships)
4. Test with the Siri AI developer beta

Apple's WWDC session on App Intents highlighted several best practices:

• Cover the most common 5–10 user tasks first (not every possible action)
• Provide multiple phrase variations for each intent
• Use entity resolution to handle references like "the one from yesterday" or "my favorite playlist"
• Handle partial parameter state — Siri AI may call an intent with missing information and expect the app to prompt for it

What this means for users: Over the next 6–12 months, apps that adopt App Intents will become controllable via natural language through Siri AI. Apps that do not adopt App Intents will remain accessible only through their own interfaces.

Siri AI vs ChatGPT vs Gemini

Users searching for "Siri AI" are often comparing Apple's approach against the AI assistants they already know. Here is how they differ.

The practical difference comes down to ecosystem depth versus breadth. Siri AI is limited to Apple devices, but within that ecosystem it can do things the cross-platform assistants cannot — reading your messages, creating calendar events, controlling apps by intent. ChatGPT and Gemini work across more devices, but they operate as standalone apps with no deep system integration.

For a user who lives entirely inside Apple's ecosystem, Siri AI is more capable because of the system access. For a user on multiple platforms or who wants the latest model capabilities (reasoning, coding, image generation) on demand, ChatGPT or Gemini remain the better choice for those specific tasks.

What Is Still Missing or In Development

The WWDC 2026 announcement was substantial, but the developer documentation also makes clear what is not ready yet.

Onscreen awareness is listed as "in development" in the current Siri developer docs. The demo during the keynote showed the feature working, but it may not be present in the initial public release. Apple's pattern with major Siri features — like the original SiriKit, or Apple Intelligence itself — has been to announce at WWDC and ship key features later in the OS lifecycle.

Third-party app actions depend on developers shipping App Intents schemas. While Apple has built the framework, the real-world capability depends on adoption across the App Store. Early adopters include major categories — messaging, ride-sharing, productivity, media — but the full ecosystem will take time.

Multi-step compound tasks — asking Siri AI to do something like "find the recipe Alex sent me, add the ingredients to my shopping list, and order anything I do not have" — work in controlled demos, but Apple still marks parts of this experience as subject to future software updates. Treat complex cross-app chains as an area that may change through the beta period.

On-device model size tradeoffs remain a constraint. While Apple's neural engine is powerful, on-device LLMs are smaller than cloud-based counterparts. For tasks that require broad knowledge or complex reasoning, Siri AI may defer to Private Cloud Compute or deliver a simplified response. Apple has not published benchmark comparisons against GPT or Gemini.

Frequently Asked Questions

Does Siri AI require an internet connection?

No. Many Siri AI features work entirely on-device — personal context queries, conversational interactions, onscreen awareness. Requests that exceed on-device compute capacity will require an internet connection for Private Cloud Compute. Apple states this is a minority of requests.

How is Siri AI different from the Siri in iOS 18?

Siri AI runs on fundamentally different architecture — LLM-based reasoning instead of fixed intent matching. Previous Siri updates added features (more languages, proactive suggestions, SiriKit), but the core engine remained the same. Siri AI is a complete replacement of that engine.

Will Siri AI work with CarPlay?

Yes. Apple confirmed that Siri AI will support CarPlay. The conversational improvements — maintaining context across turns, handling interruptions — are especially relevant for driving scenarios where hands-free interaction matters most.

Can I opt out of Siri AI and use the old Siri?

Apple has not announced a way to revert to the previous Siri engine. The old Siri architecture is deprecated. Users who disable Apple Intelligence entirely may receive a reduced version of Siri, but the new engine is the default for supported devices.

How do developers start preparing?

Developers can download the Xcode 27 beta and start defining App Intents schemas. Apple has published an updated App Intents framework reference, WWDC session videos, and sample projects. The most important step is identifying the 5–10 most common user tasks in the app and modeling them as App Intents.

Will Siri AI support HomeKit and smart home devices?

Yes. HomeKit and Matter integration remains, and the conversational improvements make smart home control more natural — "set the living room lights to warm and turn on the TV" works as a single sentence.

Does Siri AI work with Apple Watch without an iPhone nearby?

Watch models with the S9 chip or later can process some Siri AI requests on-device. More complex requests will relay to the paired iPhone or, if the iPhone is not available, to Private Cloud Compute.

Summary

Siri AI is the most significant change to Apple's assistant in its 15-year history. The rebrand from Siri to Siri AI reflects a real architectural shift: from fixed intent matching to LLM-based reasoning, from single-turn commands to conversational context, from system-only controls to third-party app actions via App Intents.

What Apple has shown so far already demonstrates the conversational improvements that are meant to make Siri feel fundamentally different to use. What is still coming — onscreen awareness, full app action coverage, reliable multi-step task execution — will define whether Siri AI meets the potential shown in the keynote.

Three things to watch:

1. September 2026 — the general release with iOS 27 will determine which features actually ship
2. Developer adoption — Siri AI is only as powerful as the apps that adopt App Intents
3. Model iteration — Apple's foundation models will improve, but whether they close the gap with GPT and Gemini depends on how aggressively Apple invests in on-device model research

If you are on an iPhone 16 Pro or later, put the public beta on your calendar for July. When it launches, enroll your device at beta.apple.com, install the iOS 27 beta profile, and within the first day ask Siri AI one question you would have asked the old Siri — the difference in response quality will be immediately obvious. The deeper features — personal context, onscreen awareness, app actions — will arrive through the iOS 27 lifecycle starting in September.

Article updated June 9, 2026. Based on WWDC 2026 announcements, Apple Developer documentation, and Apple Intelligence preview pages. Features and timelines are subject to change before general release.