I started as a hospitalist at a community hospital and was asked, two years in, to lead the hospital medicine section. We were four physicians covering a hospital that needed far more than four physicians could give. Building that group taught me the lesson that has shaped everything since: the limiting factor in medicine is rarely knowledge or effort. It’s the system the knowledge and effort have to pass through. We grew the group fourfold, built one of the first collaborative physician and APP models in our region, and put clinicians on night shifts in ways nobody around us was doing. The model outlived my tenure, which is the only kind of success that counts.

Around the same time, our hospital was implementing an electronic health record, and I made the mistake of being the physician who was curious about it. That curiosity became a career. I learned that an EHR implementation is not a technology project. It’s a trust project. That work grew from one hospital to dozens, from medical director to associate CMIO, and eventually to leading clinical decision support for a large health system, where my favorite work was subtraction: retiring hundreds of thousands of alerts that interrupted clinicians without helping patients.

The night that changed how I think about all of it came from a sepsis prediction model our team deployed. The algorithm flagged a patient of mine as septic. I had examined that patient. I had not seen it. We acted, and the patient lived. I carry that night as evidence of what becomes possible when systems are built to catch what tired, busy, human clinicians will inevitably miss. That program was later associated with a twenty percent reduction in sepsis mortality and a national award.

The next chapter took me away from the hospital floor and into innovation leadership, where I spent years learning what technology looks like from the other side of the table. We built a computer vision program to prevent patient falls from the ground up. We were among the earliest adopters of ambient listening. We formed partnerships across pharmacy and revenue cycle. Along the way I got an education you cannot get in a hospital: what it takes to turn an idea into a product, and a product into something that survives contact with a real clinical unit. I also learned how often that journey fails when the people building the technology have never spent a night taking care of patients.

That period taught me the risks as clearly as the promise. I watched models drift, saw how bias hides in data, and came to understand that the same tools capable of catching the sepsis I missed are equally capable of causing harm at scale if deployed carelessly. It’s why I founded an AI Center of Excellence during those years, and why I arrived at the sentence that now anchors most of what I write and speak about: the hard part of AI in medicine is never the model. It’s the people and the process around it.

I still advise technology companies large and small on building products that work in the real world of medicine. But most of my focus has returned to operations: leading hospital medicine and the AI portfolio that serves it across a large health system, while still practicing as a hospitalist. After years of building tools, I wanted to stand where they land and be accountable for whether they actually help. The question I ask of every technology, every workflow, every initiative is the same one I started with fifteen years ago: does this help the person taking care of the patient in front of them.