--- source_url: "https://www.thelec.net/news/articleView.html?idxno=11187" title: "[Apple's AI Strategy①] \"Our Siri Has Changed\" < IT·Gaming < 기사본문 - The Elec Inc." mirrored_at: 2026-08-10T03:34:36.185Z host: www.thelec.net cited_in_42a: true mirror_canonical: "https://index.42a.ai/www.thelec.net/news/articleView.html__q__idxno_11187" --- > **Original source:** https://www.thelec.net/news/articleView.html?idxno=11187 Artificial intelligence was the clear centerpiece of Apple's Worldwide Developers Conference (WWDC) 2026. Apple entered the generative AI race in 2024 with the launch of Apple Intelligence, but delays to several core features fueled both expectations and concerns. In particular, the next-generation Siri — featuring personal context, screen awareness and cross-app integration — had become a symbolic challenge for Apple's AI strategy after repeated postponements. At this year's WWDC, which began on June 8, Apple once again put Siri's AI capabilities front and center. By strengthening personal context, screen awareness and app integration, Apple evolved Siri from a voice assistant into something closer to an AI agent. Personal context refers to customized information and contextual understanding that AI retains from interactions with users. Apple also expanded Visual Intelligence, which understands camera and on-screen information, by integrating it into the camera app and extending support across platforms including the iPad and Mac. The company further introduced third-generation Apple Foundation Models with enhanced reasoning and multimodal capabilities, while upgrading the Apple Intelligence platform to extend AI functionality across the operating system. While general users focused on the new Siri AI, developers and the broader AI industry paid closer attention to Apple's expansion strategy for its Foundation Models Framework. By opening the framework not only to Apple's own AI models but also to a wider range of AI services, Apple signaled ambitions to build the underlying platform where AI operates. **○ Siri Finally Becomes an Action-Oriented AI** The new Siri AI unveiled at WWDC 2026 represents more than a feature upgrade. It is effectively an entirely different product. Since its debut in 2011, Siri pioneered the mainstream adoption of voice assistants. But as the generative AI era accelerated, Siri increasingly came to be viewed as lagging behind competing services. It remained limited to answering questions or performing simple commands, with constrained abilities to understand conversational context or execute complex tasks across multiple apps. ![Siri AI provides answers to questions related to on-screen content and improves personal context understanding through cross-app search. (Source: Apple)](https://cdn.thelec.net/news/photo/202606/11187_11211_1621.jpg) Siri AI provides answers to questions related to on-screen content and improves personal context understanding through cross-app search. (Source: Apple) The new Siri AI evolved into an AI agent capable of understanding user information and performing real tasks across apps. The biggest change is its "personal context" capability. Previously, Siri's understanding of user data was highly limited. Requests such as "Find the file Manager Kim sent me last month" were effectively impossible to handle because users themselves had to remember whether the file was stored in email, messages or the Files app. The new Siri AI can now understand context across emails, messages, schedules, notes, files and photos. When users search for documents or reference past conversations, Siri AI simultaneously scans multiple data sources to retrieve relevant information. This marks a shift away from keyword-based search toward intent-based understanding. Apple said Siri AI will now function as a personal assistant capable of understanding users' broader digital activity. Overseas media outlet The Verge described the update as a major transformation centered on personal context and cross-app integration. Another major addition is "Screen Awareness." Previously, Siri could not properly understand what users were viewing on-screen. Even when information was visibly displayed, Siri could not utilize it, forcing users to explain the context separately. Siri AI can now recognize information displayed on-screen and use it to perform actions. For example, if a friend sends an address via text message, users can simply say, "Add this address to the contact," and Siri AI will understand the displayed content before completing the task. ![Siri AI recognizes on-screen content to provide nutritional information and descriptions. (Source: Apple)](https://cdn.thelec.net/news/photo/202606/11187_11212_178.jpg) Siri AI recognizes on-screen content to provide nutritional information and descriptions. (Source: Apple) Apple first demonstrated this capability at WWDC 2024, but the feature was never released due to conflicts between Siri's legacy architecture and the new AI structure. At WWDC 2026, Apple finally implemented the scenario in a working environment. The capability goes beyond basic screen reading by understanding the context of what users are doing. This effectively brings "context awareness," considered a core requirement in the generative AI era, into Siri AI. Tom's Guide described the upgrade positively, writing that "Apple finally fixed Siri." Perhaps the most significant change is Siri AI's cross-app orchestration capability. Previously, Siri could only execute limited functions within individual apps. Even when users issued voice commands, they often still had to open apps and manually finish tasks themselves. Siri AI can now move across apps to complete workflows. Users simply describe their intended result, and Siri AI automatically handles the necessary steps across Photos, Messages, Mail and Calendar. Apple demonstrated Siri AI searching for photos containing specific people taken at a certain location, adding them to a shared album and sending them to family members. Users only specify the desired outcome while Siri AI performs the underlying search, album management and message delivery tasks. The workflow closely resembles an AI agent capable of acting on behalf of users. Instead of manually operating separate apps, users simply describe goals while Siri AI executes the process. Business Insider described the update as Apple giving Siri an "AI glow-up." ![](https://cdn.thelec.net/news/photo/202606/11187_11213_1758.png) Siri AI's conversational ability has also improved significantly. Earlier versions frequently lost context when users slightly changed wording or asked follow-up questions. For example, after asking "What's the weather in Seoul?" Siri often failed to understand a follow-up such as "What about tomorrow?" The new Siri AI is designed to retain conversational context and understand user intent even when wording changes or corrections are made. The experience moves closer to the natural conversational interactions offered by modern generative AI systems. Apple also emphasized that users no longer need to phrase commands in rigid formats. Unlike earlier versions of Siri that depended on predefined sentence structures, the new system can interpret more natural language. Users are meant to feel less like they are interacting with a machine and more like they are speaking with a real assistant. Siri has long been cited as one of the clearest examples of Apple lagging in the generative AI race. Some overseas media outlets argued Apple fell behind after delaying the Siri AI features it first unveiled in 2024 for more than a year. The Siri AI introduced at WWDC 2026 is the result of a two-year redesign effort. More importantly, it reflects Apple's push to transform Siri into an AI hub connecting the broader Apple operating system ecosystem. Where the previous Siri functioned as a voice interface performing isolated tasks, Siri AI now operates more like an AI agent that understands intent and coordinates apps and data to achieve goals. Users can also leverage generative AI features through Siri AI, including document summarization, email drafting and long-message organization. Siri AI ultimately operates as the most accessible interface into the broader Apple Intelligence platform. **○ Foundation Models Framework Expands Into AI Platform Layer** While Siri AI attracted the most attention at WWDC 2026, developers focused more closely on the large-scale expansion of the Foundation Models Framework. Apple first introduced the Foundation Models Framework at WWDC 2025, allowing external developers to integrate Apple Intelligence foundation models into their apps. At the time, Apple's message was straightforward: developers would be able to build AI functionality directly on-device without relying on external AI servers or APIs. One year later, Apple expanded the framework beyond an on-device AI API into a broader platform. With the addition of multimodal image support, server-model integration, Python SDK support and external AI model connectivity, the Foundation Models Framework is evolving into the core infrastructure layer of Apple's AI ecosystem rather than remaining a simple Apple Intelligence invocation tool. At WWDC 2025, the framework's defining feature was on-device execution. Developers could use Apple AI models running directly on iPhones and Macs without building GPU server infrastructure or paying external AI API fees. This was especially attractive to startups and smaller developers. At WWDC 2026, Apple took the strategy further. The central shift is toward enabling developers to use any AI model through a unified structure. The most important addition is multimodal support. While last year's framework focused primarily on text, the 2026 version expanded to multimodal environments capable of understanding image inputs, aligning Apple with broader generative AI market trends. ![The Apple Foundation Models Framework has begun supporting multimodal inputs including images. (Source: Apple)](https://cdn.thelec.net/news/photo/202606/11187_11214_1851.jpg) The Apple Foundation Models Framework has begun supporting multimodal inputs including images. (Source: Apple) Developers can now feed user photos, screenshots and documents into AI models directly within apps and utilize the resulting analysis. The move also ties closely into Apple's broader Visual Intelligence strategy. Industry observers are also paying close attention to server-model support. Until now, Apple's AI strategy largely centered on on-device AI, with the company emphasizing privacy and efficiency advantages through local computation. But the rapid evolution of generative AI has exposed the practical limitations of purely on-device approaches. Complex reasoning, large-document analysis and advanced agent tasks are difficult to process using compact on-device models. Apple's newly added server-model support appears intended to address those constraints. Developers can now choose between local device models and server-based models depending on performance requirements. Apple effectively signaled a shift away from a strict on-device-only strategy toward a more pragmatic hybrid approach. Perhaps the most symbolic change in the framework expansion is support for external AI models. Apple has begun providing an abstraction layer designed to prevent developers from becoming locked into specific AI models. Developers can now select models including Anthropic Claude and Google Gemini in addition to Apple's own offerings. For Apple, historically regarded as one of the industry's most closed ecosystems, the move is highly unusual. ![The Apple Foundation Models Framework is seeking to attract a broader range of external developers through the addition of a Python SDK and Linux environment support. (Source: Apple)](https://cdn.thelec.net/news/photo/202606/11187_11215_1926.jpg) The Apple Foundation Models Framework is seeking to attract a broader range of external developers through the addition of a Python SDK and Linux environment support. (Source: Apple) The addition of a Python SDK also carries strategic significance. The SDK supports Linux environments, marking a notable departure from Apple's traditional developer ecosystem centered around Swift, Xcode and macOS. The mainstream generative AI ecosystem today largely revolves around Python and Linux. Through the new SDK, Apple appears to be attempting to attract machine learning researchers and AI engineers — not just existing iOS developers — into the Foundation Models Framework ecosystem. The shift suggests Apple is broadening its overall developer strategy for the AI era. Rather than competing directly against GPT or Gemini at the model level, Apple appears focused on transforming its massive installed ecosystem of iPhones, iPads and Macs into a comprehensive AI platform. Siri AI serves as the user interface layer, Apple Intelligence functions as the operating system-wide AI layer, and the Foundation Models Framework acts as the developer ecosystem foundation. As a platform technology capable of affecting millions of apps and tens of millions of developers, the Foundation Models Framework also represents the infrastructure underpinning Apple's transition toward becoming a broader AI platform operator connecting multiple AI models, developers and services.