Many people trying to develop software treat AI as magic pixie dust they can sprinkle on a problem to make it go away. For coding, this as "AI-led engineering." The literature is rife with the failures of this.

Instead, I argue that we need to bring back well-known engineering best practices and treat AI as a tool to support them. This is "engineering-led AI." I created a free, online tutorial to explain this approach. A year and a half in development, but created as a happy accident of research, this approach to software development with AI requires experienced developers and leads to code that, while not perfect, is better than what I find engineering teams in many companies produce.

My background: I started programming in 1982, programming in many different languages, and am currently working with 225+ developers to help them improve their technical maturity, including with AI-assisted coding.

And for anyone who asks, my personal setup is Claude Code with Superpowers, Ponytail, PAAD (https://github.com/Ovid/paad), and some personal skills. My professional setup is Kiro with PAAD and some personal skills and powers.