What makes AI work in a hospital, and what makes it harder - Dragan Shahpaski

Many AI systems in healthcare are auxiliary tools running on top of products built for people, with a model added later. They operate on top of a systems that were never designed to be operated by anything but humans. The alternative is to assume from the start that a machine will read the system and act on it which changes the data model, the API surface, and where decisions sit in a workflow. This talk is about the shift from AI layered on top of healthcare software to healthcare software designed from the ground up for AI agents.