Ambient clinical AI that keeps the doctor looking at the patient
Healthcare AI that captures what happens in a visit and carries it into the medical record, so clinicians stop choosing between being present and being accurate.
The problem
A clinical visit produces far more signal than ever reaches the medical record. The physician is either present with the patient or accurate in the notes, and the system quietly asks them to choose. What gets lost is not only detail. It is the context that makes the next clinician’s decision a good one, and it disappears in the gap between the conversation and the chart.
Every attempt to close that gap before this one had asked the clinician to do more work: another template, another field, another few minutes after the door closes. That is why they failed. A documentation tool that costs a physician thirty seconds per visit will not be adopted, however good the underlying technology is.
The work
I came in near the founding of the company and owned delivery: turning a clinical AI thesis into something that holds up in a real exam room, on real hardware, inside workflows that clinicians are not going to change for the sake of software.
Healthcare is unforgiving about that last part. The constraint is not model quality. It is that the product has to disappear into a routine that was set long before it arrived, and earn its place without asking anyone to learn anything. Most of the difficult decisions were about what the product would refuse to do, rather than what it could.
Where it stands
Document Medical is live and the work is ongoing.