Apple Releases Siri AI, the Largest Overhaul Since Siri's Inception
Apple officially launched Siri AI, marking the largest overhaul since Siri's inception and the most significant update for Apple Intelligence to date, rolled out alongside updates for iOS 27, iPadOS 27, and macOS 27.
The core change in the new Siri is its ability to understand users' personal data—previously, Siri could only answer general questions or execute simple commands, while the new Siri can search users' emails, messages, and photos across applications, linking information scattered across different apps. A typical scenario provided by Apple is: when a user asks what to do when family visits, Siri can find the conversation record "let's bake a family recipe" from messages, retrieve the specific recipe previously sent by a relative from emails, and automatically add the required ingredients to the shopping list in reminders. This cross-application information integration capability was completely unattainable by previous versions of Siri.
The new Siri also introduces "onscreen awareness," enabling it to understand the content displayed on the user's current screen and act accordingly; the "Visual Intelligence" feature has expanded from supporting only iPhone to the entire platform including iPad, Mac, and Vision Pro, allowing users to ask Siri questions directly using screen captures or camera feeds; a writing tool has been embedded within Siri, enabling it to assist in drafting or editing content in any text input box, even mimicking the user's communication style for specific contacts when writing emails and messages.
Apple has also created a standalone app for Siri, with conversation records synchronized across devices via iCloud, allowing users to start a conversation on iPhone and seamlessly continue on Mac. This design signals that Apple no longer positions Siri merely as a shortcut entry point but aims to develop it into a sustainable conversational AI assistant; on the Mac, Siri AI is directly integrated into the Spotlight search box, which can be used to ask questions directly.
The Apple Watch has received a new set of features called "Audio Intelligence," requiring Series 12 or Ultra 4 models; the most notable is the Live Rewind feature, which allows users to double-click the digital crown to convert the last 15 seconds of conversation into text in real-time, particularly useful for users with hearing impairments or those who lose focus during meetings; the accompanying Siri Recap feature will automatically generate summaries and key points after a conversation, although these two features will only be available in a testing version by the end of 2026 and initially will not support the EU region. Additionally, Image Playground has added realistic image generation capabilities, Safari can automatically organize tabs by theme and create custom extensions based on descriptions, and shortcuts now support creating complex automation processes using natural language descriptions.
In terms of product competition, Siri AI is Apple's direct response to the competitive wave of AI assistants like ChatGPT and Google Gemini, with its biggest differentiating advantage being the deep data integration capability within the Apple ecosystem—able to access personal data from all native apps, which no third-party AI assistant can currently achieve. However, Apple has set clear limitations on hardware requirements (iPhone must be from the 16 series or 15 Pro/Pro Max, iPad and Mac must have M1 chip or above) and language support (initially supporting only English, with French, Japanese, Korean, Portuguese, and Spanish to be added next month, while Chinese is not included in the support plan and has no timeline). Additionally, due to regulatory requirements, Siri AI and other new Apple Intelligence features cannot be used in mainland China for the time being, meaning that the user base able to fully experience these features in the short term will be concentrated in English-speaking markets, where the user stickiness effect within the Apple ecosystem will first be realized, while the experience upgrade for Chinese users and the mainland market will lag significantly.
Source: Public Information
ABAB AI Insight
Apple's historical progress in the AI assistant field has been relatively slow and bumpy—Siri debuted with the iPhone 4s in 2011 as the industry's first mainstream voice assistant, but its functionality has evolved slowly over the past decade, lagging behind Amazon's Alexa and Google Assistant; in 2024, Apple first announced its "Apple Intelligence" strategy at WWDC, originally planning to bring significant personalization and cross-App understanding capabilities to Siri by 2025, but related features have repeatedly been delayed, with reports of postponements due to technical immaturity. The launch of Siri AI is, to some extent, Apple's effort to fulfill its previous delayed promises.
To catch up with the generative AI wave, Apple has previously partnered with OpenAI to integrate ChatGPT into Siri for handling some complex queries, and there have been reports of considering incorporating Google Gemini to provide underlying large model capabilities for Siri; this means that Apple is pursuing both internal research and external large model procurement to bridge the generational gap with OpenAI and Google in generative AI experiences, while leveraging the vast user data accumulated within the Apple ecosystem (emails, messages, photos, etc.) as a competitive asset.
This strategy aligns with Microsoft's deep integration of Copilot into the Windows and Office ecosystems and Google's embedding of Gemini into the Android and Workspace ecosystems—tech giants are embedding generative AI capabilities into their operating systems and native application ecosystems to leverage data and distribution channel advantages against "bare model" vendors like OpenAI; however, Apple's differentiated path emphasizes "device-side data integration" capabilities rather than merely competing on the general intelligence level of models. The industry is currently at a critical window where "operating system vendors are using ecosystem data to counterattack general large model companies."
This essentially represents a shift in pricing power: the core competitiveness of AI assistants is shifting from "underlying large model capabilities" to "platform data integration capabilities." The mechanism behind this shift is that as the gap in mainstream large model capabilities narrows, relying solely on the intelligence level of models is no longer a decisive barrier; whoever can compliantly and securely access users' cross-App and cross-device personal data and transform it into a usable assistant experience will regain initiative in user stickiness and product differentiation. This is why, despite falling behind in general large model capabilities, Apple can still rebuild its moat in this round of AI assistant competition through its ecosystem data integration capabilities.
ABAB News · Cognitive Law
Models compete in IQ, ecosystems compete in memory.
Where the data resides, the moat lies.
The weapon of the laggards' counterattack is always their existing user base.