Modern AI models are optimized to predict the most likely answer—which is exactly why they often produce surprisingly average work. This talk explores how to build AI systems that deliberately resist that tendency. Rather than relying on a single frontier model, we’ll treat creativity as a distributed systems problem: assigning specialized cognitive roles to different models, introducing controlled randomness, building feedback loops, and using critique and iteration to evolve ideas instead of accepting the first plausible response. Through a real-world autonomous content pipeline, we’ll explore how concepts from DevOps—modularity, orchestration, observability, and resilience—apply just as naturally to creative AI workflows. The result isn’t just better content; it’s a framework for building AI systems that consistently generate ideas you wouldn’t have thought of yourself.