The Chart Doesn’t Listen

Notes on two kinds of medicine, and what patients keep trying to tell us

By Luis Cisneros, CEO of Serelora

Spend enough time around medicine in different parts of the world and you notice something strange. The rooms with the least in them often hold the most of what medicine is actually for.

Medicine is intimate in a way almost no other work is allowed to be. Strangers undress for you. They sit on crinkling paper in a gown that closes in the back, cold, half-naked, frightened, and they tell you things they have never told their spouse. That is the raw material of the profession. Everything else, the imaging, the labs, the codes, exists in service of that moment or it exists for nothing.

The American exam room is the best-equipped room in the history of the profession. Imaging that borders on science fiction. Labs back before lunch. A specialist for every organ and, if you are patient, a sub-specialist for every corner of the organ. It is a marvel and I mean that without irony. If something has gone seriously wrong inside your body, there is no better place on earth to be.

And in the middle of all that capability sits a physician with their back half-turned, typing.

Nobody starts out that way. That physician arrived wanting to heal people, the oldest reason there is, and then spent a decade being trained, drilled, and audited into a second profession they never applied for. Record the encounter. Justify the diagnosis. Support the billing code. Satisfy the regulator. Build a record sturdy enough to survive a courtroom, because someday it might have to. They do this beautifully, by the way. The American medical note is a genuine achievement, airtight and defensible, a small legal fortress built fresh forty times a day. Nobody dreams of building forty small legal fortresses a day. They do it because the system pays for what it can measure, and it can measure documentation. Listening leaves no trace. And what leaves no trace, no system on earth has figured out how to reward.

So the doctor types. Dutifully, skillfully, often at eleven at night, dinner gone cold, family asleep, finishing the day’s notes in the blue light of a screen. The research has a name for this, pajama time, which may be the saddest euphemism in all of healthcare. There is a famous finding that it takes thirty-two clicks to order and record a flu shot. Thirty-two. A flu shot. Somewhere in this country tonight, a physician who once held a human heart in their hands is fighting a dropdown menu.

Now go south. In much of Latin America the physician works with less. Fewer specialists, older machines, a pharmacy that stocks what it stocks. The waiting room is plastic chairs and a ceiling fan pushing warm air around, and some of the people in those chairs rode two hours of bad road to sit in them. What the doctor has to offer, reliably, sometimes exclusively, is the conversation. The visit that starts with your mother’s health before it gets anywhere near yours. The doctor who knows your neighborhood, your work, what the water is like where you live. I watched this kind of medicine up close for years, and those doctors were no saints, believe me. What they had was attention, the one clinical resource that costs nothing, which is why systems with nothing run on it, and why systems with everything forget they ever had it.

Attention turns out to be a diagnostic instrument, maybe the original one. Patients do not hand over their story like a lab slip. They hedge. They trail off. They joke about the thing that terrifies them. They mention the chest pain on the way out the door, hand already on the knob, the medical equivalent of a confession shouted from a departing train. The most important fact in the room is usually the one nobody says out loud, and either a human being catches it or nobody does.

Here somebody raises a hand. We measure this, they say. Satisfaction scores are strong. Communication metrics are fine. Sure. Jeff Bezos, a man with more dashboards than any human alive, kept reading the angry emails customers sent him anyway, and he explained why. “When the anecdotes and the data disagree, the anecdotes are usually right.” The dashboard says the encounter is working. Maybe the problem is the dashboard.

Because look at what American patients are actually doing. They are walking out of the most capable medical system ever assembled and paying cash, real money, to functional medicine practitioners, naturopaths, and chiropractors. About the merchandise, I will be honest. A lot of it is expensive theater, the evidence behind much of it runs from thin to indefensible, and I suspect plenty of the customers quietly know it. They go anyway. They go because somebody in a calm room spends an hour asking about their sleep, their stress, their marriage, their whole aching life, and nods, and looks at them the entire time. They walk out with something the correct diagnosis and the correct prescription somehow did not deliver. The experience of being heard.

That exodus is the anecdote. It contradicts the metric. The metric is wrong.

None of this is a verdict on individual doctors, and it should not be read as one. Plenty of American physicians listen like confessors, and many of them do it by donating the time out of their own evenings, which is its own quiet scandal. Systems produce behavior the people inside them would personally disown. Most Americans carry no racial hatred in their hearts, and systemic racism grinds on regardless, because systems answer to incentives and incentives have never once checked what was in anybody’s heart. The American physician wants to listen. You can tell, because they keep trying to do it while typing.

Which brings us to the machines, and to my own work, because I build this technology and I have no intention of pretending otherwise. Ambient AI can now capture an entire clinical encounter. Every word, structured, summarized, coded, filed. This is real, and it matters. But let us be honest about what got automated. The transcript is complete. The hesitation before the answer is not in it. The fear underneath the casual question is not in it. The thing the patient circled three times and never quite landed on, not in it. A machine can capture what was said. Understanding what was meant remains stubbornly, gloriously, a human job.

So the people building this technology face a choice, and it is the whole ballgame. Build AI that helps physicians feed the record faster, and you have built a nicer keyboard. The screen keeps its seat between two people, where it has sat for thirty years. Or build AI that takes the record off the physician’s hands entirely, and you give back what three decades of health IT quietly borrowed and never returned. The freedom to turn around and face the patient. Doctors have been asking for exactly this the whole time. Nobody becomes a physician to type.

Charlie Munger liked to say, in his flat Omaha way, that you have to see the world as it is rather than as you would prefer it, because problems become solvable the moment you quit flattering yourself. So let us quit. Patients are leaving the best-resourced medicine on the planet to buy an hour of eye contact. That is the market talking, and the market, whatever else you think of it, does not lie about where people spend their own money.

Medicine gets measured by how completely you understand the person sitting in front of you, cold and frightened, on the crinkling paper.

The chart is just what is left over afterward. The residue. Nobody ever got better from a note.