“Sorry, surgery isn’t an option because you won’t have a good outcome. Get radiation therapy. Goodbye.”
That, in essence, is what a surgeon at a nationally recognized cancer center told a healthcare industry veteran whose advanced cancer had come back. The appointment, says the patient, lasted three minutes.
The patient didn’t accept it and went on to see half a dozen physicians and finally found a team at another major cancer center that gave the patient confidence, not just because of its expertise, but because everyone there, from the receptionist to the nurses to the physicians, understood the patient’s fears, and was committed to helping.
The story has a happy ending, too. The patient had surgery, the option that supposedly wasn’t open, and now reports no complications and excellent test results.
This Isn’t an AI Story. Yet.
Let me be clear: This patient’s story has nothing to do with artificial intelligence. No algorithm told the first surgeon to wrap up the appointment in three minutes. And the patient draws an independent lesson, which is about access: finding the right expertise shouldn’t depend on having the knowledge and connections of a healthcare insider.
But I see the patient’s story with a different thought in mind: A three-minute appointment is what you get when a system treats throughput as the product. And AI is the most powerful throughput tool medicine has ever been handed.
The Minutes Are Real. Sort Of.
Ambient AI scribes are the first AI tool most physicians have actually lived with, and the evidence on what they do is coming into focus. According to a May 2026 MGMA summary of the research, randomized and multicenter studies show they reduce burnout and documentation burden. The productivity story is murkier: estimated time savings of roughly 7 to 22 minutes per provider per day, and one longitudinal study found only about a 2 percent rise in billed wRVUs over 150 days. An editorial in NEJM AI summed it up in its title: “AI scribes are not productivity tools yet.”
Note the word “yet.”
Somebody is going to make those minutes productive. The only question is who gets to define “productive.”
Three Places the Minutes Can Go
First, the saved time can go into more visits. That’s the spreadsheet answer. At the high end, twenty minutes a day is roughly one more patient per physician, every day, forever. You can see why an investor-backed MSO or a hospital with productivity targets would want those minutes. (Last week I wrote about the Stop Corporate Takeovers of Physicians Act, which targets entities that dictate how long a physician may spend with a patient. Congress has noticed.)
Or, second, the saved time can go into the chart. The MGMA flags that some apparent AI gains may reflect higher coding intensity rather than more care. That’s the kind of pattern that attracts payer scrutiny. Those minutes come back with interest, in an audit.
Or, third, the saved time can go into the patient. A few more minutes to answer the question the patient was afraid to ask. A nurse who returns the patient’s call. A physician who says “let’s get you another opinion” instead of “you won’t have a good outcome.”
Of the three options, only the third one builds anything.
Why This Is Your Group’s Edge
Here’s where this thinking gets practical. In a corporate-owned practice, someone else decides where the minutes go, and that someone answers to investors. In a physician-owned group, your physicians decide, through tools you control: your compensation plan, your governance, your AI-use policy, and the agreements you sign with hospitals and vendors.
For more than a decade, I’ve written about the Experience Monopoly™: delivering an experience for patients, referral sources and facilities that competitors can’t replicate. That patient we started this post about didn’t choose that team because it was efficient, but because everyone, starting with the receptionist, treated the patient like a person. As AI makes the technical side of medicine faster, cheaper and more uniform, the experience is the part that won’t commoditize, unless you let it.
Think of it this way: Efficiency is a tool; it isn’t the product.
Some Timely Takeaways for You
- Decide where the minutes go before someone decides for you. Put it on your board’s agenda. If AI is saving your physicians time, adopt an explicit position on what that time is for.
- Look hard at your compensation plan. A pure unit-based production formula has already answered the question: more visits. If you want time with patients to count, the plan has to say so.
- Read your hospital and vendor agreements. Look for productivity standards, visit-time expectations and performance metrics tied to tools someone else selected. A hospital that buys the tool may expect the throughput.
- Audit AI-assisted documentation before a payer does. Faster notes that drift toward higher codes aren’t savings; they’re exposure.
- Measure what you want to keep. If you track only RVUs, you’ll get RVUs. Track patient experience and outcomes, too.
That three-minute appointment didn’t need AI. But AI can make the next one shorter, or it can buy back the time that made the difference for that patient. Make sure your group is the one deciding which.
If you’d like to talk through what this means for your group, let me know.
Related posts


