Jason Williamson

Jason Williamson is the CEO of MythWorks, where he leads the development of deterministic, verifiable AI that runs efficiently at the edge. A former leader of startup and ecosystem programs at AWS and Oracle, he brings enterprise rigor to scaling new technology and partners. Jason has taught data science at the University of Virginia, served as a venture partner advising founders on product and go-to-market, and earlier helped build analytics capabilities in financial services. A U.S. Marine Corps veteran, he applies mission-first discipline to ethical AI, including efforts to combat human trafficking. His work centers on turning AI into measurable business outcomes with clear ownership, governance, and KPIs.

Smaller Models, Bigger Wins

Episode Summary

Jason Williamson pulls back the curtain on efficient intelligence and why raw scale is not a strategy. He shows how MythWorks squeezes serious reasoning from tiny compute, placing deterministic, verifiable models where work actually happens, from factory lines and fleets to robots and field devices. Instead of token prediction, you get right or not solvable, with every step logged for audit. Assistants are tuned to the job to be done, so a plant engineer gets line-change guidance, a route planner gets validated paths, and FP&A gets variance explanations that reconcile to the penny. Agents operate like a real workforce with named owners, access rights, SLAs, and change control that your operations and compliance leaders will sign. You will learn the efficiency math that beats GPU sprawl, how to design for managed autonomy so humans keep decision rights, and the data patterns that travel well to the edge. We cover approval gates for sensitive actions, rollback plans, data residency, and red team drills that build trust before scale. Most important, we get concrete on measurement: time to decision, cost per inference, energy per task, defect escape rate, and revenue per employee. This is AI you can run, trust, and afford.