Healthcare AI Trust Starts With Data, Not Models

Healthcare AI is moving from pilots to production, raising urgent questions about clinical trustworthiness. Snowflake's Jesse Cugliotta argues the foundation must be built in data infrastructure before any model is trained, citing the challenge of 46,000-word patient charts that physicians cannot realistically review. Komodo Health's Marmot platform addresses this by aggregating over 330 million de-identified patient journeys into a unified Healthcare Map built on Snowflake. CTO Amit Sangani emphasizes that every analytic step is logged, auditable, and reproducible, giving healthcare providers full transparency into how AI-generated answers are reached.