Never gets tired.
Five stories underground in Wisconsin, an 82-year-old woman explained to twenty thousand of us what she thinks we're still for.
This is the last essay in my three-part series on Epic Systems and Judy Faulkner. Part one was the founding story. Part two was the succession plan. Same disclosure as the first two: I was at their annual meeting because I work in Epic every day. Nobody paid me. Nobody asked for any of this.
This last one is personal, because it's about my job.
What she said
Underneath all the state fair decorations, this year's meeting was really about one thing: AI. Epic has something like 160 AI projects going. Software that preps the chart before the visit. Software that writes the note while you talk to the patient. Agents that do the paperwork of medicine. A research database of hundreds of millions of de-identified records that can flag which patient is probably coming back through the ER doors.
At some point Judy laid out how she sees the division of labor. Humans bring judgment, common sense, empathy, and intuition, which she described as instinct merged with intellect. The AI brings pattern recognition across enormous data sets. It iterates fast. It's always available.
And it never gets tired.
Twenty thousand people nodded. I nodded too. I've been tired in ways I wouldn't wish on anyone, and a machine that never fades sounded like mercy.
It wasn't until the flight home that I figured out what was bugging me.
Some history first
I should say the quiet part, because every physician reading this is already thinking it.
The electronic chart is one of the great achievements of modern medicine and also one of the main reasons my colleagues are burned out, and both of those are substantially Judy Faulkner's doing. Ask a room of doctors what's grinding them down and the chart comes up before money, before hours, before almost anything. We spend roughly two hours in the software for every hour with a patient. Her software.
So yes, there was something strange about watching the company that built my documentation burden promise that its newest products will hand me back the attention its older products took. A friend of mine at the conference put it in one word: arsonists. Selling extinguishers.
Why he's wrong
Because the tools work. Or at least, they're starting to.
Epic's ambient charting is already live in hundreds of health systems. Nurses using it say they're getting time back at the bedside, which is the entire point of nursing. There's radiology AI moving cancer patients into treatment faster. If a piece of software can listen to my visit and write an accurate note while I keep my eyes on an actual human face, I want it yesterday. I'm a tech nerd. I've never pretended otherwise.
And I have no patience for the doctors doing loud, performative resistance to all of it, like refusing a good tool is some kind of virtue. It isn't. It's usually just pride with a stethoscope on.
So no, my worry isn't the machines showing up.
My worry is that one line. The one we all nodded at.
Tired is information
Never gets tired sounds like a pure advantage until you've spent enough years around actual sick people.
Fatigue is a signal. The unease you can't name at two in the morning. The sense that this airway is going to be harder than it looks. The feeling that a patient's story isn't adding up even though every number on the screen is fine. Judy called that intuition, and she's right. What she didn't say is where it comes from. It comes from a body. It's built out of thousands of nights, some of them bad, most of them long, all of them felt. The machine is tireless because nothing is at stake for it.
The machine never gets tired because it has nothing to lose. Care comes from the one in the room who does.
To her credit, Judy said the humans can't come out of the loop, and I believe she means it. My concern is quieter than robot doctors. Tools shape the people who use them. If the software preps the visit, drafts the assessment, and suggests the plan, then the easiest possible version of my job is to become an approval step. Click accept, next patient. Nobody will ever decide that doctors should stop thinking. It'll just get very, very smooth to stop, and tired people take the smooth path. I know because I'm one of them.
The bottom line
Where I landed, after three weeks of chewing on this giant:
Take the tools. All of them. Let the machine write the note, prep the chart, fight the insurance company. Every minute it hands back belongs at the bedside, and refusing that trade to make a point spends your patients' attention, not yours.
But the trade only pays off if we actually spend those minutes on the patient, and no software update can do that part. The machine will never get tired. Fine. Then maybe the whole job now is that we can be tired and notice anyway. That was the work back when the chart was paper, and I don't think any release date changes it.
Usual caveat: I could be wrong about how this plays out. The tools are young, I'm one anesthesiologist in Kentucky, and the people building this stuff are smarter than me about software. They are not smarter than you about your own attention.
So here's my question for you this week: where in your work has a tool quietly turned you into an approval step, and what would it take to start noticing again?
Reply and tell me.
That's all for this week.
See you next Saturday.