Physician and FDA advisor Shantanu Nundy joins Stanford hosts Matt Lungren and Justin Norden for this installment of Stanford's AI in Healthcare Series, discussing how hospitals and regulators keep tabs on AI tools once they are deployed in real clinical settings. The conversation covers the FDA's evolving approach to regulating adaptive AI systems, the problem of alert fatigue among clinicians overwhelmed by algorithmic warnings, and why robust post-deployment monitoring and feedback loops matter as much as initial validation. Nundy draws on his experience advising the FDA to explain the gap between how AI tools perform in trials and how they behave once doctors and nurses actually use them day to day. The three speakers trade examples of what effective tracking looks like in practice and what happens when it is missing, framing patient safety as an ongoing operational challenge rather than a one-time approval checkpoint. It runs as a conversational format rather than a formal lecture, but the content stays focused and substantive throughout.