Coding Intelligence for Beginners Group Lab

A Podkey summary of Coding Intelligence for Beginners Group Lab, from WWDC 2026.

App development is starting to feel very different. Instead of jumping straight into code, developers are increasingly using agents to turn rough ideas into specs, code, tests, fixes, and sometimes even a full working feature in one extended loop. A lot of the shift comes down to speed and cost, but the more interesting part is control. In Xcode especially, these agents are getting woven into the actual shape of the project, the testing workflow, the undo history, and even the security boundaries around what they can touch.

• From vague idea to working feature
• Why spec-driven work matters
• Cheaper experiments, bigger swings
• Tests become the guardrails
• What Xcode adds that generic AI tools don't
• Agent mode versus chat mode
• Control, permissions, and visibility
• Undo, rollback, and learning from the agent
• Where agents still mess up
• Local models, cloud models, and privacy
• Working offline and staying current

This podcast was created with Podkey. Make your own at https://podkey.fm