Updates

With over 110,000 satisfaction responses from more than 165,000 patient interactions, we explore what measuring acceptability carefully has taught us about building healthcare AI that truly serves patients.

Anyone with access to Claude can build a healthcare chatbot in a weekend. But getting from a demo to a regulated product clinicians can actually use with their patients can still take months or years of work. Here are the things I believe are critical to get adoption and are difficult to shortcut.

Six years of real-world research and more than 150,000 patient conversations later, we have shown that there is a better way to deliver healthcare.

We wouldn't license a physician who could diagnose accurately but couldn't communicate with patients. So why deploy AI systems that way? Our new Nature Medicine paper proposes a framework for evaluating clinical AI - drawing on lessons from autonomous vehicles, that goes beyond diagnostic accuracy to assess both 'cure' and 'care' behaviors.

In the week before Christmas, a patient at Dell Medical School at the University of Texas in Austin completed a routine follow up after eye surgery. The conversation was led by Dora, Ufonia's AI clinical assistant—marking the first US trial call.