Apple adds new AI “foundation models” capabilities and pushes Xcode upgrades aimed at faster app building

Apple used the June 8 opening of its Worldwide Developers Conference to outline a new push for AI-assisted software creation, positioning developer tooling as the next lever for faster iteration on iPhone, iPad, Mac, and other Apple platforms. The company also made Xcode 27 beta available to developers the same day, indicating how quickly it expects teams to start testing the new workflow.

In its developer-facing materials, Apple described Xcode as moving beyond code suggestions into “agentic” development, where coding agents help drive work toward a developer’s goals. The company said Xcode 27 includes coding agents, a Device Hub feature intended to manage devices from one place, updates to performance and testing tools, and localization enhancements aimed at reducing the friction of shipping in multiple languages.

Apple’s Xcode overview also emphasized how “coding intelligence” is integrated into the IDE. It said predictive code completion is powered by an on-device machine learning model trained for Swift and Apple SDKs, and that developers can interact with code using the large language model of their choice. The company also pointed to “Coding Tools,” which it described as IDE features that can help write code and documentation and fix errors from within the editor.

For preview and simulation workflows, Apple highlighted capabilities in its Xcode product description that allow developers to visualize changes while building, and to prototype across devices using Simulator. MacRumors, covering the WWDC 2026 announcement, said Apple’s coding assistant was expanded to handle app localization and to interact with simulated devices, with additional details expected in later sessions.

On the AI framework side, Apple’s messaging centered on expanding the Foundation Models framework, which gives developers access to the on-device “foundation model” at the core of Apple Intelligence. Apple previously said the framework is designed to support intelligent experiences that can run offline, with AI inference described as free of cost, and that it is built for privacy-oriented deployment on-device.

Newer WWDC coverage pointed to additional capabilities layered onto that foundation. MacRumors reported that the Foundation Models framework gains image input support, letting developers pass images alongside text into on-device models, and that Apple introduced custom skills and server-side model execution as part of the framework. MacRumors also reported that Apple announced a new Core AI framework alongside these Foundation Models changes.

The same reporting also tied Apple’s AI tooling to broader app integration through App Intents, describing an expanded support surface for letting users ask Siri to perform actions inside third-party apps. App Intents, in plain terms, are a framework for exposing what an app can do (and with what parameters) so system features like Siri and Shortcuts can call into those actions.

Apple did not provide, in the accessible materials summarized here, a full technical breakdown of the new intelligence framework, including which server-executed models will be supported, what performance or latency targets Apple expects, or how the new image-and-text workflow will interact with privacy controls and developer compliance requirements across different jurisdictions. Apple also did not disclose specific timelines for when every developer-tool enhancement would be available beyond the Xcode beta availability and the general descriptions of capabilities.