Oh boy, another AI talk! No, really! Let's focus on architecture and coding decisions that lead to systems that are better-suited to extension and maintenance by AI agents. Many of you have experienced the speed with which agents can spin up small greenfield applications, while simultaneously noting how this is often not the case with existing legacy software. Why? What is it about legacy software that makes it so difficult for agents to move as quickly there as they do when creating a proof-of-concept web app? In this session we'll answer these questions and identify pragmatic and concrete actions we can take to improve the maintainability of our apps with an eye toward making them more malleable by AI agents. And, it turns out, this also makes the systems more maintainable by human programmers, too.