Herding AI Centaurs: Managing the AI-Native Development Team
Details
For 15 years, the best chess players were "centaur" players, humans who played with a computer and could beat either humans or computers. That is where we are for software development and technical teams.
But how to work well with an AI is not easy or obvious, especially at the team level. AI can help developers produce code faster, but it can also flood teams with slop pull requests, turn off developers' brains, burn thousands of dollars in tokens, and turn you into a hyperactive army of botsitters.
This talk is about managing the system around AI-assisted development, not just what model is the most popular this week. We will examine how fast feedback, strict engineering guardrails, and deliberate code reviews can turn AI-generated output into reliable software.
You will learn how to manage reluctant adopters, prevent overreliance on AI, and get management buy-in. AI can improve delivery rather than simply increasing the volume of code.