You can just ask AI to do things

Sean Goedecke, a GitHub engineer, hands AI agents his routine engineering chores but keeps his own messages, design decisions and code review, because he answers for them.

Sean Goedecke, a software engineer who works on GitHub Copilot, GitHub's AI coding assistant, lists the everyday tasks he now gives to AI agents. He argues that most engineers are held back less by what AI models can do than by noticing which tasks agents could take on. He also names the work he keeps for himself and explains why.

• His AI agents start his development server and run through the main user paths with test requests instead of him doing it by hand.
• When he deploys or changes a feature flag, an agent checks metrics and logs before and after, and can keep checking every hour for the rest of the day.
• At 6am a scheduled agent reads his unread messages, flags the questions and problems and looks up relevant logs. His American colleagues start work hours before him in Australia.
• He writes his own messages and reports, makes his own design decisions and reviews code himself, and usually takes up none or one of an agent's suggested tasks.

An engineer can list repeated chores and hand them to an AI agent while keeping the decisions they are accountable for. The author works on GitHub Copilot and says so; the practices he describes work with other agent tools too.

Derived by Working Surface from the article; more in the Playbook. Source line: I don’t ask the agent to do everything.

30 October 2026: as AI agents do more of the building, teams decide on purpose which work people keep and set up tests, usage figures and baselines before handing over more.