Is Siri finally going to become an AI assistant? The situation is like this. At Apple's 2026 WWDC, a series of Siri updates were released, all focusing on AI functions, including context, screen awareness, and cross-App task execution. In addition, Apple has released more Apple Intelligence functions, including picture editing, passwords, text messages, Safari... and so on.
It can be said that as long as it is an Apple-native App, it must be more or less related to Apple Intelligence.
Moreover, this WWDC is also Cook’s last developer conference as CEO.
But surprisingly, John Ternus, the next Apple CEO, did not take the stage to give a speech.
Okay, let’s take a look at what updates Apple has brought to WWDC this time.
Siri
Siri is the protagonist of the entire WWDC 2026.
Craig Federighi, senior vice president of software engineering, calls Siri AI a "entirely new version", a new version powered by Apple Intelligence.
Apple’s official press release describes it this way: Siri is now a more capable, more conversational assistant with personal contextual understanding, extensive world knowledge and screen awareness.
Translated into human language, Siri in the past was like a voice remote control that could only execute commands. This time Apple wants to turn it into a system-level agent that truly understands you.
The specific updates are divided into three layers.
The first level is personal context.
Siri AI can now extract information from your emails, text messages, photos, notes, and calendars.
If you ask it "What is the name of the restaurant recommended by my friend last week?", it will go through your Messages history, find the conversation, and tell you the name of the restaurant; if you ask it "What is my hotel confirmation number?", it will go to Mail to find the reservation confirmation email and extract the number; if you ask it "the photos taken on vacation last month," it will understand the time range and photo content, and directly call up the photo album.
Siri is now not only upgraded for searching, it can also connect these isolated data to form a complete memory.
The second layer is screen awareness.
Siri AI can see what you're currently looking at, understand the information on the screen, and take action based on that information.
If you receive a message in Messages, "Can you bring a dish for Saturday's dinner party?", you can directly say to Siri, "Help me think of what dish to bring, and then add the recipe to Notes."
Siri will understand the context of the text message, give you suggestions, and then actually create a new note in Notes.
When you see an address in Safari, you can say "Add this address to the contact card" and Siri will recognize the text on the screen, find the corresponding contact, and complete the update.
The third level is execution ability.
Siri AI can now perform tasks across apps.
You can ask it to "lighten this photo and send it to Pedro." It will open the photo editing tool, adjust the exposure, save it, then open Messages, find Pedro's conversation, and send the photo. During the entire process, you don’t need to manually switch apps or click any buttons.
You can ask it to "remind me to call Vicki at 3pm tomorrow" and it will create a reminder, set the time, and associate the contact. Based on this, you can ask it to "Add this event to calendar" and it will start parsing the event information, creating a schedule, filling in the location and time.
Together, these three capabilities turn Siri from a tool that only answers a single question into an AI agent.
The most picturesque scene is the Siri mode in the iPhone camera.
Apple moved Visual Intelligence from last year’s Camera Control button to the Camera App, turning it into an independent mode alongside photos, videos, portraits, and panoramas.
You switch to Siri mode, point the camera at the food in front of you, and directly ask "How many calories does this dish have?"
Siri will identify the food, query nutritional information, and give you detailed ingredient and calorie data.
If you ask a plant "What kind of flower is this?" it will tell you the variety, maintenance methods, and flowering period. You ask a poster "When will this event start?" It will recognize the text, extract the date, and even add it directly to the calendar for you.
Moreover, on Mac and iPad, Siri has more entrances. You can invoke it directly from Spotlight, select "Ask Siri" from the right-click menu, and enter questions in any text box.
In Vision Pro, Siri can even appear in space as a 3D virtual object. You look at it and speak, and it will understand the focus of your sight, know which object you are looking at, and then give relevant information.
And Apple has also launched an independent Siri App.
This app can save and sync your conversation history. You can look through previously asked questions, continue previous topics, and ask Siri to remember what you talked about.
As early as 2011, Siri debuted with the iPhone 4S, and its demonstration effect was amazing.
You can ask about the weather, set alarms, and send text messages. Voice interaction was a sci-fi experience at the time.
But users soon discovered that Siri could understand the instructions, but not the context. You ask it "What's good to eat nearby?" and it gives you a list of restaurants. But if you then ask "Which one has the highest rating?" it doesn't know that you are talking about the list just now. You have to say the complete question again.
In 2016, Apple announced at WWDC that it would open SiriKit to third parties, allowing developers to connect their apps to Siri.
However, the thunder is loud and the rain is small. SiriKit's capabilities are strictly limited to a few categories, such as messaging, payment, taxi hailing, fitness, and photo search. There is very little that developers can do.
More importantly, Siri's own understanding ability has not changed qualitatively. Third-party access only gives it a few more app names to call out. In essence, it is still a voice assistant that cannot truly understand your intentions.
At WWDC in 2024, Apple promised to rebuild Siri with Apple Intelligence to give it personal context understanding, screen awareness, and cross-App mobility capabilities.
However, this commitment was delayed in 2025. Siri itself has changed little over the course of a year. Therefore, people in the industry continue to ask, "Has Apple fallen behind in the AI era?"
In the past fifteen years, Siri has experienced too many "this time is really different" moments, but every time users get it in their hands, they will find that it is still the same voice assistant that cannot understand complex questions.
Therefore, the WWDC in 2026 is actually very stressful for Apple. They need to prove that they can not only make AI, but also put AI into Siri, a product that has existed for fifteen years, and make it truly a usable intelligent assistant.
Apple Intelligence
The key to why Siri can have such a big upgrade lies in Apple Intelligence.
As early as 2024 WWDC, Apple Intelligence has already appeared.
Apple told developers and users that we have our own AI framework, which can do Writing Tools, generate pictures, and summarize emails. But at that time, Apple Intelligence existed more as an independent function. You needed to actively call it, click that button, and open that menu.
The wait has come to an end.
Craig repeatedly emphasized one word on stage, deeply integrated. Deep integration.
He said that Apple Intelligence is no longer a separate chat box, but is hidden in daily scenes such as photos, Safari, passwords, text messages, phone calls, calendars, and family. Users do not need to "turn on AI", AI will automatically appear in what you are doing.
A common problem with AI products in the past two years is that users need to change their usage habits to adapt to AI.
You have to remember to open the specific app, type in the questions you want to ask, and then copy the answers it gives you.
However, what Apple wants to do is to do the opposite, adapt AI to the user's existing usage habits, and let it appear naturally in the apps you already use.
Photos App is one of the focuses of this update.
Apple has launched Spatial Reframing, a feature that uses AI and 3D modeling technology to generate new perspectives from existing photos.
During the demonstration, Apple engineers showed a group photo. The original composition was a bit tight. After using Spatial Reframing, the perspective of the photo changed, as if you took a step back and took another photo, and the composition became more comfortable. The engineer said, “It’s like I was able to go back in time and adjust my camera in the moment.”
Original picture
Reframing
The Extend function can extend the photo boundary, adjust the proportion, and straighten the horizon. You take a photo of a seascape and the horizon is crooked and part of the sky is cropped off. Extend can help you fill in the sky and straighten the horizon, making the photo look like it was taken with a better composition.
The Clean Up function has been upgraded and can now handle more complex scenarios. You take a portrait with passers-by, trash cans, and telephone poles in the background. Clean Up can remove these debris and do it very naturally without leaving any obvious traces of AI retouching.
Image Playground now supports generating photo-realistic images. In the past, Image Playground could only generate illustration-style or animation-style pictures. This time Apple added a photorealistic style, which can generate images that look like real photos. You can use it to generate lock screen wallpapers, generate contact avatars, and even modify images based on existing photos.
And Apple also stated at WWDC that all photos edited with AI will have a hidden SynthID watermark.
This is a technology jointly developed by Apple and Google. It can embed invisible marks at the pixel level without affecting the quality of the image, proving that the image has been processed by AI. This is Apple’s response to the “AI picture authenticity issue.”
In an era where AI can generate photos that look fake and real, Apple chose to tag each AI-edited photo so that users and the platform can identify which pictures were originally taken and which ones have been altered by AI.
In addition to the picture function, updates to Apple Intelligence are also reflected in Safari.
Safari can now automatically organize tabs.
You have opened dozens of web pages, some for shopping, some for searching information, and some for work, all piled in the tab bar.
Safari’s AI can identify the content of these web pages and automatically group them according to categories, grouping shopping websites together, grouping work documents together, and grouping news articles together. You can also use natural language to describe the classification method you want, such as "put all web pages about coffee together", and Safari will understand your intention and complete the classification.
The Notify Me function can monitor web page changes. You are waiting for a product to be restocked, or waiting for a price to drop to a psychological price, or paying attention to the terms update of a website.
You can ask Safari to monitor this page and send you push notifications when the content changes.
Although this function seems a bit "crude", it solves a real user need. Especially when you buy tickets and grab seats, you don't need to refresh the web page repeatedly or set up third-party monitoring tools. Safari will keep an eye on it for you.
You can even describe the Safari extension you want in natural language, and AI will generate it for you.
However, this feature is still in its early stages.
Passwords App has also been upgraded with Apple Intelligence.
Passwords can now not only tell you which passwords are too weak or have been compromised, it can also automatically fix them for you.
You click "Fix" and Passwords will call Safari, automatically visit the website, log in to your account, find the page to change the password, generate a strong password, complete the change, and save the new password.
You don’t need to do anything manually during the entire process. Passwords will “agentically take action on your behalf” and act on your behalf agent-like.
Messages can now give one-click suggestions based on the content of the conversation. You receive a message "Don't forget to have a meeting at 2pm tomorrow", and Messages will display a button "Add to Reminders" below. You click and the reminder is created, with the time and content automatically filled in. You receive a message saying "This is my new home address" and Messages will suggest "Add to Notes" or "Update Contact".
The Smart Reply feature in Mail and Messages can now learn your personal writing style. Smart replies in the past were all generic templates, such as "Okay", "Received", and "Thank you".
The new Smart Reply will analyze how you usually write emails and reply to messages, and then generate reply suggestions that match your tone. If you usually write more formal emails, it will give you formal advice. If you usually use emojis a lot, it will also add emojis to the suggestions.
The Call Context function of Phone App is a very practical little innovation.
For example, you need to call the airline customer service and provide your reservation number, flight number, and identity information.
Now Phone App can automatically find relevant information from Mail and Messages and display it on the screen when you make a call. While you are on the phone, you can see the confirmation number and booking details, and read them directly to the customer service.
The most easily overlooked is the update of Shortcuts.
Shortcuts has always been the most powerful but the most difficult feature on iOS. It allows you to automate almost any operation, but the prerequisite is that you have to learn how to build processes, how to set trigger conditions, and how to call APIs.
For most ordinary users, Shortcuts is a tool that “looks great but I can’t learn how to use it.”
Now Apple has added the “Describe a Shortcut” feature. You describe what you want to automate in natural language, such as "When I leave the company, send Pedro a message saying I'm on the way with an estimated time of arrival."
Shortcuts will understand your needs, automatically build this process, set geofence triggers, call the Messages API, obtain navigation data, and generate message templates. You just need to confirm and the automation is set.
If this function is really useful, it may turn Shortcuts from a "geek toy" into an automated entry point for ordinary people. Because it lowers the barrier to a minimum, you don't need to learn any technical concepts, you just need to explain clearly what you want to do.
Cook's curtain call
This year, Apple’s liquid glass design has caused quite a bit of controversy.
Many users feel that the new frosted glass effect is too heavy, the interface elements are not clear enough, and the visual hierarchy is confusing.
At this WWDC, Apple quietly made adjustments. They provide more transparency and visual effect adjustment options, allowing users to adjust the contrast, clarity, and motion intensity of the interface according to their own preferences.
Performance optimization is a reserved item at WWDC every year, but this year Apple talked about it in more detail than in previous years.
They said that photo display speed has been improved, AirDrop transfer is faster, CPU scheduling is more efficient, and multi-tasking switching is smoother.
Since Apple has added a lot of AI content this time, and it is all processed locally, this requires stronger computing power.
So while Apple is talking about AI, it is also talking about performance, optimization, and basic experience.
At the same time, Apple also announced at WWDC that personal contextual capabilities can be extended to third-party apps, but only if developers access capabilities such as Spotlight and App Intents.
Siri AI can understand the content in third-party apps and perform operations in third-party apps, but all this must be actively adapted by developers.
In fact, Apple users spend most of their time in third-party apps, such as WeChat, Douyin, Taobao, Meituan, Netflix, and Spotify.
If Siri AI cannot understand the content in these apps and cannot help you do things in these apps, then it is still an assistant that can only control system functions and cannot be regarded as a true system-level agent.
Apple provides developers with the Foundation Models framework, App Intents API, Writing Tools integration, and Image Playground integration. They say the tools are free, have no per-request fees, have built-in privacy protections and all processing can be done on the device.
But whether developers will buy it depends on the actual results.
This is a chicken-and-egg problem. Users find Siri AI difficult to use because third-party apps are not connected. Third-party apps are not connected because users don’t use Siri AI very much.
Apple needs to use actual data to prove that the usage rate of Siri AI is growing in the next few months, and prove that developers who access it can obtain user growth or increase in usage time. Only in this way can the third-party ecosystem really take off.
Finally, it’s Cook’s curtain call.
Cook will transition to executive chairman on September 1, 2026, and Ternus, senior vice president of hardware engineering, will take over as CEO. This means that WWDC 2026 is Cook’s last developer conference as CEO.
When Cook took over as CEO in 2011, Apple had just lost Steve Jobs. At that time, there were widespread doubts about whether a supply chain expert could lead a company known for innovation.
But it took Cook fifteen years to turn Apple into the world's most valuable company, with revenue growing from US$108.2 billion in 2011 to more than US$400 billion in 2025, and the market value once exceeded US$4 trillion.
He transformed Apple from a hardware company into a hardware plus services company.
Apple Music, Apple TV+, Apple Arcade, iCloud, and Apple Pay, these service businesses grew from scratch during Cook’s tenure and are now an important pillar of Apple’s revenue. He promoted the birth of Apple Watch, AirPods, and Vision Pro, and expanded Apple's product line from Mac and iPhone to wearable devices and spatial computing.
AI is clearly the biggest headache at the end of Cook's tenure.
When Apple launched Apple Intelligence in 2024, the features were incomplete, Siri's promises were not fulfilled, and the entire AI strategy looked rushed. The extension to 2025 has made the outside world question whether Apple has fallen behind in the AI era.
Therefore, WWDC in 2026, for Cook, is not only the last developer conference he will host as CEO, but also a critical moment when he needs to prove that Apple's AI strategy has not failed.
At the end of the press conference, Cook said, "It's been the honor of a lifetime." He thanked the developers and the team, saying "I truly believe the best is still ahead." I truly believe the best is still ahead. He said Apple's North Star has always been to create products that serve people's needs.
"Polaris" is a metaphor, meaning the long-term unchanging direction, standard of judgment, and highest guidance. In this sentence, it means that no matter whether Apple makes hardware, software, or AI, it will ultimately come back to one question: Is this thing really useful to users? Does it make people’s lives and work better?
As mentioned earlier, Ternus, as the incoming CEO, did not give a speech during the entire press conference.
Everyone thought that at this WWDC, Cook would give him a "pass the torch" moment and let him come on stage to say a few words and introduce the future direction. But Ternus never showed up from beginning to end.
This may be because WWDC is a software-focused event and Ternus' background is in hardware engineering.
He joined Apple in 2001 and participated in the development of iPhone, iPad, AirPods, and Mac product lines, and led Apple's transformation to Apple Silicon. But he is not a software person, nor is he an AI expert. It may not be the most appropriate time for him to talk about AI strategy at WWDC.
Therefore, Ternus’ first show is likely to be the iPhone launch event in September. After all, that's what he's best at.