In this episode of the "Training Data" podcast, Pat Grady interviews Daniel Nadler, co-founder of OpenEvidence, an AI copilot trained on peer-reviewed medical literature designed to assist doctors in making informed decisions. Nadler discusses OpenEvidence's consumer-like adoption by over 100,000 physicians, contrasting it with the traditional, slow-moving healthcare technology adoption process. He attributes this success to focusing on doctors as consumers with real pain points, specifically the overwhelming influx of new medical information. Nadler explains how OpenEvidence helps doctors quickly find answers to edge cases by indexing peer-reviewed medical literature, including an exclusive partnership with the New England Journal of Medicine. He also emphasizes the importance of "openness," ensuring accessibility for all doctors, including those in underserved communities. Nadler details the AI architecture, highlighting the use of smaller, specialized models trained on curated medical data to minimize hallucinations and the importance of infrastructure and high-IQ, neuroplasticity individuals in building successful AI applications.
