Why the real AI moat lives in your data when models are commodities

Terry Miller, Vice President AI & Machine Learning at Omada Health, calls frontier models "very expensive commodities" and argues the real moat lives in proprietary data and the workflows built on top of it. With 14 years of longitudinal data, Terry shares how his team built an agentic harness using reusable templates, why tightly bounded applications succeed where broad ones fail, and what "edge healthcare" means for the future of decentralized care. Chapters 00:00 | AI in Healthcare Today 03:14 | The $5 Trillion Problem 06:50 | Demo to Production Lessons 08:09 | LLMs Are Pattern Matching 09:50 | Productivity vs. Production Workflows 12:37 | Building the Agentic Harness 17:57 | Traditional ML Meets LLMs 20:06 | Data as the Real Moat 25:27 | Edge Healthcare From Home 32:55 | AI Policy and Regulation