The Craft of Care.

Medicine is changing course, the way a river does when the ground shifts underneath it, and AI is moving that ground faster than anything before it. Much of what’s being automated right now is built for speed and volume, and it shows in tools that produce more notes, more messages, and more output without much regard for whether any of it is good.

I think the better path is collaboration between clinicians and AI with craft built in. Craft is a form of care, and care is a form of craft. Both depend on attention, judgment, and a willingness to get the details right for one particular person, which are exactly the things most at risk when technology is designed without clinicians in the room.

If you’ve sat with a family after a hard diagnosis, or tried to fit a whole life into a fifteen-minute visit, you know things that never show up in a dataset. That knowledge belongs in the rooms where these tools are being designed, and Anastomosis exists to help more clinicians bring it there.

The craft of care shows up in specific places. These are some of the first I'm working through, and the list will keep growing as the work does.

01 The Medical Record

Clinical notes started as a way for clinicians to talk to each other about a patient. Over time they turned into billing instruments, liability documents, and compliance artifacts, and the patient mostly fell out of them. AI scribes are now writing notes at enormous scale, which is genuinely useful, but if the logic underneath documentation stays the same, we'll just be automating the wrong thing faster. Bringing craft back to the record means treating it as an account of a human encounter first and a billing artifact second.

02 The Clinical Conversation

Medicine has never been especially good at talking with patients, between the jargon, the rushed visits, and the assumptions about how much someone can take in after frightening news. Patients now turn to chatbots between visits and ask language models to explain their lab results, which tells us less about patients than about what they weren't getting from us. The craft here is redesigning how clinicians communicate, with technology strengthening the relationship instead of standing in for it.

03 The Care Environment

The spaces where care happens shape the care itself, from the lighting and acoustics of an exam room to whether a waiting area feels urgent or calm. My training in landscape architecture taught me to see the built environment as something that acts on people, which in medicine makes it a clinical variable. The craft is in designing physical and digital spaces around what patients need in order to heal, rather than around what's easiest to staff or cheapest to build.

04 Clinical Decision-Making

AI is already shaping who gets a hospital bed, which diagnoses get flagged, and whose symptoms carry more weight in a triage algorithm, and most of the clinicians using these tools had no part in building them. Judgment is the oldest craft in medicine, and part of practicing it now means questioning an algorithm the way you'd question a surprising lab value, while making sure the people shaping these systems remember what was lost the last time medicine optimized for something without asking what it was optimizing toward.

05 Patient AI and Clinical AI

Health AI isn't one thing. Some tools are built for patients, like chatbots that explain lab results or help someone prepare for a visit, while others are built for clinicians, like scribes, decision support, and inbox triage, and a growing number try to serve both at once. Each kind carries different assumptions about who's using it, what they already know, and who's accountable when something goes wrong, and those differences rarely get examined. Doing this with craft means designing each tool for the person who will actually use it, and thinking carefully about how they all fit together around the same patient.

06 Learning the Craft

Clinicians have always learned their craft through repetition, feedback, and time alongside people who are better at it than they are. AI is changing that for medical students who now study with it, for residents who train alongside it, and for experienced physicians catching up on tools that didn't exist when they finished training. The real question is how to use AI to deepen clinical skill without quietly replacing the practice that builds it, and how to make sure the people building these tools learn enough about clinical work to build them well.

This is the work, and there's more of it than any one list can hold. If you want to be part of it, start here.

See the coaching program