Post by Sean Kamkar

At Zest AI's IQ AI Lending Forum, I got to sit down with Omar Hatamleh, Ph.D — former Chief AI Officer at NASA and one of the sharper thinkers I've encountered on what it actually takes to scale AI in complex organizations. The conversation was genuinely fun, and a few things stuck with me. 1. Scaling AI is an operational problem, not a technology one. Organizations won't fail because the intelligence isn't good enough — they fail because the infrastructure, processes, and internal alignment weren't built to support systems that have to keep learning. 2. Trust is still the binding constraint. The most powerful model in the world doesn't make it into production — or stay there — without the ability to verify what it's doing, transparency into how it behaves, and continuous monitoring. That part hasn't changed. 3. Speed is redefining what risk even means. Faster decisions at greater scale shrinks the margin for error. That raises the bar on governance and resilience across the whole org, not just the model team. The pace of AI advancement means deploy-and-monitor isn't a strategy — you have to be built to iterate as fast as the technology does. Standing still is falling behind. Good week. Grateful for the conversations. Returning recharged.