Before we can reliably delegate the work to the AI agents, we need them to understand the project structure, the architectural patterns in use, and to have a sense of how important different things are. In this session, I will show what works for me. You will see which types of tasks I use the GitHub Copilot agent for locally in Visual Studio or VS Code, and when I delegate tasks to run in the background on GitHub or in Azure DevOps.

Now, when my project is ready for AI agents, let's build a real-world feature live on the stream. I will implement a couple of things in our new SaaS product and explain how I think about various aspects of the process. Shall we provide detailed specifications, or is it better to let the AI agent come up with their own ideas? How important are tests and human code review, and how can we ensure we are not increasing technical debt?

I'll answer all the questions you post during the show.