AI is everywhere, but how do we know if it's actually good? The teams winning with AI aren't just moving fast, they're measuring well. And that measurement starts with a human-centered definition of "good" that lives in design and research. AI evals have long been an engineering problem. That's changing. As capabilities commoditize, design is the new moat. This talk is an invitation for designers and researchers to lead and become architects of how we evaluate AI against real human needs.