Apple laid out its most ambitious set of artificial intelligence changes in years during the Worldwide Developers Conference last week. The announcements stretch from a rebuilt Siri now called Siri AI to new frameworks that let coders tap powerful models without building their own infrastructure. Executives positioned the moves as practical advances rooted in privacy. Yet they also reveal how closely Apple now works with Google to deliver what users expect from modern assistants.
Craig Federighi, Apple’s senior vice president of software engineering, described the new Siri as “a dramatically more capable and conversational assistant designed to help users find information and get things done throughout the day.” The quote comes from Apple’s official newsroom release. Federighi highlighted personal context understanding, onscreen awareness and access to broad world knowledge that pulls up-to-date answers from the web. Siri can now scan a user’s messages, emails and photos to surface a hotel confirmation from last year or recommend a restaurant mentioned by a friend. It can look at what is on the screen and suggest adding a recipe from a text thread directly into the Notes app.
But. The foundation underneath these abilities relies on a partnership with Google. Apple’s next-generation foundation models were built in collaboration with Google’s Gemini technology, according to multiple reports including CNET’s coverage published yesterday. The arrangement lets Apple keep data processing private while borrowing the scale of models it did not develop entirely in-house. Private Cloud Compute handles requests without storing personal information. Apple says outside experts can verify this promise at any time.
A dedicated Siri app arrives with the changes. Conversations sync privately across iPhone, iPad, Mac, Watch and Vision Pro through iCloud. Users can revisit old threads or start new ones in one place. Visual Intelligence expands too. On the iPhone it integrates into the Camera so a user can point, tap the shutter and ask Siri to split a bill using Apple Cash or analyze nutrition in a dish. The feature reaches iPad via screenshots, Mac through a keyboard shortcut and Vision Pro by simply looking at an object.
Writing tools sit inside Mail, Messages and third-party apps. Siri can generate a draft from a description, adjust tone to match a manager’s preference for short bullets, offer suggestions and proofread automatically. Dictation gains accuracy, automatic formatting and customizable voice expressiveness. These details echo reporting in TechCrunch’s comprehensive recap from Monday.
Developers receive even more direct help. The Foundation Models framework now offers free access to Apple’s models on Private Cloud Compute for any developer with fewer than two million first-time App Store downloads. That eliminates infrastructure bills for smaller teams. The framework adds image input support, lets coders call third-party models such as Claude or Gemini through the same Swift API, and introduces Dynamic Profiles that manage multi-agent workflows. Apple plans to open-source the framework later this summer. The information draws directly from MacRumors’ detailed report on the Platforms State of the Union.
Core AI replaces the older Core ML approach. It gives developers tools to run custom on-device models with ahead-of-time compilation, dedicated Instruments profiling and Python utilities that convert PyTorch models for Apple silicon. The framework powers much of Siri under the hood while keeping inference local, fast and private. Susan Prescott, Apple’s vice president of Worldwide Developer Relations, said in the company’s developer-focused release, “With new intelligence frameworks and agentic coding in Xcode 27, developers have the tools they need to focus on what they do best: bringing their incredible ideas to life.”
Xcode 27 itself shrinks by 30 percent and runs only on Apple silicon. It syncs settings through iCloud, offers per-project themes and replaces the Simulator with a new Device Hub. Agentic coding takes the biggest leap. AI agents can now interact with the simulator, localize strings, run tests, pull crash reports from Organizer and fix issues autonomously. Xcode Cloud builds run up to twice as fast. These productivity gains matter because they shorten the distance between idea and shipped product. A major app like Notion is migrating its interface to SwiftUI for exactly that reason: better performance and consistency.
SwiftUI receives reorderable containers, swipe actions that work on any view, layouts that resize twice as fast and lazy state initialization. The new document infrastructure gives first-class URL access. Swift 6.4 adds an anyAppleOS availability shorthand, suppressible warnings, async support inside defer blocks and sharper type-checker diagnostics. Parts of the operating system kernel are now written in Swift, a sign of growing confidence in the language.
Design language changes arrive without escape hatches. Apps compiled with Xcode 27 adopt Liquid Glass automatically. The refined version improves content diffusion, adds darkened edges for depth perception and includes a user-controlled transparency slider. Intel Mac support ends completely. macOS Tahoe marked the last release for those chips, so developers can ship Apple-silicon-only binaries on the Mac App Store. iPhone apps running on iPad or through iPhone Mirroring now resize freely when rebuilt with the latest SDK. Some observers link the change to future foldable hardware.
Game developers gain from an updated Game Porting Toolkit that adds AI coding assistance and new Metal command-line tools. Apple’s open-source MLX framework now supports Metal 4 and can train models across multiple Macs using RDMA over Thunderbolt. A Spatial Preview framework lets Mac apps stream 3D content into the space around Vision Pro users in real time.
Performance numbers stand out. iPhone and iPad apps launch up to 30 percent faster. New photos appear 70 percent quicker. File transfers speed up. These tweaks compound the sense that Apple spent the past year tightening the entire stack. iOS 27, iPadOS 27, macOS 27 and the rest will ship this fall. Developer betas are available now. A public beta follows next month. Siri AI starts in English with more languages later. Certain features face delays in the European Union because of regulatory requirements.
Tim Cook’s final WWDC keynote as CEO carried extra weight. He is set to hand the role to hardware chief John Ternus in September. The event therefore marked both a technical milestone and a leadership transition. Analysts had expected heavy focus on Siri after years of incremental updates and reported delays. The result mixes on-device speed, cloud assistance when needed and a clear bet that context plus privacy will differentiate Apple from pure cloud competitors.
Reactions on X reflect that tension. Some developers praise the free Private Cloud Compute tier and agentic tools as genuine barriers removed. Others note the reliance on Gemini and question how much true differentiation remains. One post captured the mood: Apple’s approach to AI may stem from its historical caution, yet the outcome still delivers capabilities users have waited for. Another highlighted economic realities. Running every query in the cloud costs too much in power and servers. Edge computing becomes necessary, not optional.
Apple also opened a fifth Developer Center in Berlin this year, joining locations in Cupertino, Shanghai, Singapore and Bengaluru. The company continues to court global talent while tightening its platforms. For industry insiders the message is clear. Build with these new frameworks and Xcode agents. Ship faster. Let the models handle repetition. Focus on experiences that feel personal because they actually understand the user’s data without exposing it.
The changes do not transform computing overnight. They do, however, give developers concrete APIs, faster build cycles and smarter assistants that operate across Apple’s lineup. That combination has historically proven powerful. Whether it closes the gap with pure AI-first competitors will play out over the next year of betas, launches and app updates. For now the tools are here. The invitation to use them is open.
WebProNews is an iEntry Publication