LEAP 2026: Thinking Through Healthcare AI Workflows

At LEAP 2026 in Riyadh, I went looking at two areas: agentic AI in enterprises and AI in healthcare. What stood out on the startup floor was how many different problems sit inside that second label.

Teams were presenting work around medical imaging and treatment planning, chronic-disease management, clinical nutrition, patient access, and the operational work of medical coding and claims. These were different points in a healthcare workflow, not one interchangeable application of a model.

Some focused on decisions made by clinicians. Others addressed a patient's interaction with a service, or the administrative work that lets an organization operate. I was encountering teams and their explanations, not independently validating clinical outcomes.

My takeaway was that calling something AI-powered says much less than naming the problem it is meant to solve. Who uses the output? What decision or task comes next? Where does the output fit into the existing work?

Those questions change the engineering discussion. The appropriate integration depends on the task and the people relying on it; a model demonstration alone does not establish that the complete workflow is useful.

The clearest conversations I had came from walking the floor, asking questions, and speaking with people building those systems. I came away with a broader view of healthcare AI—and a more specific standard for discussing it: explain the workflow before celebrating the label.