Apple has unveiled a major update to Xcode—version 26.3—that fully integrates AI agents from Anthropic (Claude) and OpenAI (Codex) into its development environment. This update grants these AI systems broad autonomy to write code, build apps, run tests, and verify outputs with minimal human input, marking a significant leap toward “agentic coding.” Unlike previous AI features limited to suggestions, the new tools deeply connect to the build and test processes, allowing AI to handle coding tasks end-to-end within Xcode. Apple’s adoption of the open Model Context Protocol (MCP) enables compatibility with a variety of AI agents beyond Claude and Codex, encouraging a wider ecosystem of AI development partners. Despite past struggles with AI integration, Apple aims to ensure greater reliability through continuous build and test cycles, visual verification of changes, and the ability to roll back unwanted modifications. While AI-assisted coding promises faster development, security experts warn of risks including bugs, vulnerabilities, and the impact on open-source communities. Apple posits that deep IDE integration can serve as a quality control mechanism, though limitations remain, such as the AI’s current inability to independently debug runtime issues. The release candidate is now available to Apple Developer Program members, signaling Apple’s bold bet that agentic coding will transform software development productivity and creativity.