For the first time in a long time, it feels like Siri might actually matter again. At WWDC 2026, Apple officially retired the image of Siri as a slightly clumsy voice assistant and reintroduced it as “Siri AI” – a system-wide conversational layer powered by Apple Intelligence that is meant to sit on top of your iPhone, Mac, iPad, Apple Watch, and even Vision Pro.

This is Apple’s big swing at the AI moment: not a chatbot in a browser, but an assistant that lives everywhere in the OS, understands what is on your screen, remembers what you were doing yesterday, and claims to do all of that with Apple’s usual privacy pitch intact.

Siri, meet Siri AI

If you have been using Apple devices for a while, you know the Siri story has always been a bit awkward: launched early, hyped hard, and then slowly outpaced by everyone from Alexa to Google Assistant to the newer wave of chatbots. Apple even acknowledged that gap on stage this year, saying there were “times when you expect more from Siri” – corporate-speak for “we know it hasn’t been great.”

Siri AI is Apple’s attempt to wipe that slate clean. The company calls it “a profoundly more capable assistant,” and in practice that translates into three big upgrades: it can hold multi-turn conversations without losing the thread, it can pull in real-time information from the web, and it can act on your personal data and apps in ways that feel much closer to an AI agent than a voice command parser.

You talk to Siri AI pretty much like you talk to any modern chatbot. Apple says you can speak or type naturally, ask open-ended questions, brainstorm ideas, or just throw follow-up questions at it without constantly repeating context. In one of Apple’s keynote demos, a presenter asked about an upcoming Suki Waterhouse concert, learned that tickets required a lottery entry, and then immediately asked Siri AI to remind them when that lottery opened – without restating any details. That kind of chaining is table stakes for an LLM in 2026, but it is brand new territory for Siri.

Apple Intelligence under the hood

The “AI” in Siri AI is Apple Intelligence, the company’s homemade stack of models and system features that debuted last year and now underpins everything from writing tools to photo editing to on-device dictation. The framing is very Apple: instead of talking about parameter counts or tokens per second, Apple talks about where the models run and how tightly they are integrated with your data and apps.

On-device models live directly on supported iPhones, iPads, Macs, Apple Watch, and Vision Pro hardware, handling everyday tasks like composing messages, understanding your requests, and doing visual recognition. For heavier queries, Apple sends data to what it calls Private Cloud Compute – server-side models running on Apple silicon in the cloud, wrapped in a privacy-first story where requests are processed but not stored, and where the company insists independent experts can verify what those servers can and cannot do.

There is another twist: behind the scenes, Apple has built an AI architecture that leans on Google’s Gemini models in the cloud while still wrapping the experience as “Apple Intelligence.” In a separate briefing, Apple said it worked with Google, NVIDIA, and Intel to make sure its foundational models and Private Cloud Compute run efficiently on modern AI hardware, including NVIDIA GPUs inside Google Cloud. Apple is still selling this as “Apple AI,” but the stack is far more hybrid than the marketing suggests.

From orb to operating system

Visually, Siri AI looks different too. On iPhone, Siri now primarily lives in the Dynamic Island: swipe down from the Island, press the side button, or say “Hey Siri,” and you get a glowing Siri orb and a conversational UI that feels a lot closer to a chat app than the old full-screen overlay. Apple has given the voice a refresh as well, backed by a new engine that lets you tune expressivity and speaking rate, with granular controls over pitch, pace, and tone on supported devices.

The more interesting change is that Siri is no longer just a “voice layer” that sits on top of a few APIs. Apple is turning it into a system-wide assistant that you can summon almost anywhere. On macOS, Siri AI is integrated into Spotlight and available via right-click on files or windows, so you can, for example, right-click a PDF and ask Siri to summarize it or pull dates into Calendar. On visionOS, Siri becomes a 3D object: you can place a glowing, animated orb in your space and just look at it to start talking.

There is also a dedicated Siri app, which might be the most important part if Apple wants people to treat this as a true AI assistant rather than a one-off feature. The app keeps a scrollable history of all your Siri conversations, synced via iCloud, and lets you pin key threads so you can, say, keep a running “trip planning” or “work ideas” chat that travels with you from iPhone to iPad to Mac. It is essentially Apple’s answer to the standalone chatbot apps everyone has been installing over the last two years.

Agent behavior, not just answers

Where Siri AI starts to feel truly different is in how deeply it can reach into your apps. Apple kept stressing that this is not just about answering questions, but about taking action based on what you are doing and what is on your screen.

In one demo, a presenter asked about a dessert they remembered from an event. Siri AI dug through their Messages history to find the relevant conversation, pulled out the dish details, drafted a “watch party” menu, and then pre-filled a message to friends that included the menu, offering send and edit options. In another, Siri identified a landmark in a photo, opened navigation to that location, then surfaced photos from a previous family trip there and added a chosen image to a shared family album.

This is where the Apple Intelligence integration really shows. Siri AI can query notes, messages, emails, photos, and other on-device data to answer questions like “Show me the Wi-Fi password mom sent me last winter” or “Find that PDF contract I signed for the landlord and summarize the termination clause.” It is the same concept every AI assistant is chasing – an agent that knows your digital life – but Apple has a structural advantage in owning the OS and default apps.

The company is extending this agent behavior into smaller everyday tasks too. Siri AI can edit a message you just sent, add a song playing in the car to your workout playlist, generate a Calendar event from a natural-language description, or build a new Shortcut based on you simply describing the workflow you want. On the productivity side, Apple Intelligence can auto-group your Safari tabs into topics, monitor pages for price changes or restocks, and even help generate simple Safari extensions based on the behavior you describe.

Seeing, not just hearing

Visual Intelligence is the other pillar of this story. Siri AI is no longer limited to whatever you say; it can now look at what the camera sees or what is on your display and act on it.

On iPhone, there is a new Siri mode in the Camera app: point your camera at something – a plate of food, a storefront, a document, a cricket ball – and invoke Siri to get context, nutritional info, or smart actions like splitting a bill, saving a loyalty card to Wallet, or searching for similar items. Apple’s example was very on-brand for a global audience: the system can identify a bowl of chicken pho and surface ingredients and nutrition details with a single tap.

On Mac and iPad, Visual Intelligence lets you screenshot whatever is on-screen and then ask Siri AI about it. That might mean summarizing a dense article, extracting addresses and adding them to Contacts, or asking for more information about a product photo inside an email. On Vision Pro, Apple takes this a step further: you can simply look at objects or text in your environment and ask Siri AI questions, blurring the line between mixed reality UI and an ever-present AI lens.

It is easy to see where Apple is going: Siri as a universal “what am I looking at and what should I do with it?” layer that spans hardware categories. That is not unique in this market – Google and Samsung are doing similar things with Gemini and Galaxy AI – but Apple’s advantage again is consistency. If you are in the ecosystem, the behavior should feel largely the same across devices.

Writing, editing, and live translation

Apple is also using Siri AI and Apple Intelligence to creep into the space that tools like Grammarly, Notion AI, and chatbots have taken over: writing, editing, and communication.

A feature called “Write with Siri” lets you invoke Siri AI in most text fields across the system. You can ask it to draft from scratch (“write a friendly but firm email to my landlord about the leak”) or to revise what you have already typed for tone, clarity, or length. Apple says Siri AI can even match your writing style, punctuation, and tone in Messages and Mail, which is both powerful and a little uncanny – you might eventually see messages that look like you wrote them, but were assembled by Apple’s models.

Apple Intelligence also adds system-wide proofreading, quietly checking spelling and grammar and suggesting changes wherever you type, plus smarter suggestions in Messages and Mail that can, for example, turn a conversation into a Calendar event or a Reminder with one tap.

On the spoken side, Live Translation now stretches across Messages, FaceTime, phone calls, and even AirPods, providing automatic text and audio translations in real time. Dictation benefits from a new on-device model that is supposed to be more accurate and resilient to accents and background noise on supported devices. Taken together, these tools sketch out a future where Siri is not just a thing you talk to, but an invisible writing and language layer inside almost everything you do.

Devices, regions, and the inevitable fragmentation

Because all of this is built on Apple Intelligence, there are the usual compatibility strings attached. Apple Intelligence features, including Siri AI, run on a subset of recent hardware: iPhone 16 and later, iPhone 15 Pro models, the A19-based iPhone 17 lineup, iPads and Macs with at least an M1, newer Apple Watch models, Vision Pro with M2 or better, and a handful of upcoming devices like MacBook Neo. Some of the more advanced Siri AI voice customization options require even newer hardware – for example, iPhone 17 Pro and specific M-series iPads and Macs with at least 12GB of unified memory.

Then there is the language and geography question. Apple says Siri AI will launch later this year, in beta, in English only, and initially not in the EU because of the company’s ongoing fight with regulators over the Digital Markets Act. In a statement, Apple argued that EU rules could force it to give third-party assistants the same deep access to user data and app controls that Siri AI will have, and the company is clearly not eager to do that on day one.

For users in the US, though – the first wave Apple is clearly targeting – the message is much simpler: update your device this fall, and Siri becomes a lot more capable.

A second chance at Siri

It is hard to overstate how much of a reset this is for Apple. For years, Siri was a punchline in the AI conversation, and Apple’s strategy often looked like “wait until the dust settles, then show up late but polished.” With Siri AI, Apple is finally aligning its flagship assistant with where the rest of the industry has been heading: LLM-powered, context-aware, and tightly woven into daily workflows.

At the same time, Apple is trying to differentiate on the things it knows how to sell: deep integration, consistent design, and a heavy emphasis on privacy. Apple Intelligence’s “your data never stored” and “used only for your requests” messaging is front and center, and the company clearly hopes that positioning will resonate with people who find cloud AI exciting but unnerving.

There are still open questions. Apple has not gone into much detail on how developers will be able to hook their apps into Siri AI beyond the existing App Intents and Shortcuts frameworks, though it is promising that the same on-device models and frameworks powering Apple Intelligence will be available for third-party apps at no per-request cost. And as with any AI assistant, the real test will be less about keynote demos and more about how it behaves on a messy Tuesday afternoon when you are juggling emails, timers, kids’ schedules, and random questions about flight changes.

But after watching Siri stagnate for more than a decade, this feels different. By reimagining Siri as a true AI layer on top of Apple Intelligence – one that can see your screen, understand your context, act across apps, and remember conversations – Apple is not just catching up to the AI race, it is trying to reframe it around the devices people already live in all day.