AI and the Future of Healthcare: Balancing Technology with Human Connection

Before we start, I want to invite you to Rosetta Fest 2026 in Nashville. This is where employers, unions and clinicians who are cutting healthcare costs 20 to 50% while improving care and outcomes share exactly how they did it. Operators learning from operators with patients at the table. Learn more and register at Rosetta Fest.org. Now let's get into today's conversation. What does healthcare I do for doctors? Well, you hear about these 10x in the airs. Does the concept of a 10x doctor exist? Does that mean they're seeing 20,000 patients? If it's 20,000 patients, what does that do to a relationship? I mean, it's like way past the Dunbar number. The delta between just a general purpose, AI, what do you think is different about medical decision making that needs that kind of gap? So decision making is the operative word there. That's not really how about Pumswirk or any other. They're kind of like next word predictors in a lot of ways. The decision making part is this kind of new paradigm that needs to be created over the next decade. How does artificial intelligence think like a doctor? And that's where I think we're stopped. I would just say, be ridiculously curious. Just ride the wave. We don't know what's going to happen. All right, welcome back to another episode of Relocalizing Health, where we're exploring how we transform our healthcare system. You know, the one that taps the best of what's possible with modern approaches and technology while recognizing that far too often, we forget that health starts at home with mom and dad. It pulls out from there. But with that, I'm going to introduce our guest. He's a real digital health pioneer. He's a physician, but he's also got a public health background. And he's been thinking about healthcare transformation since coming out of residency. And to my knowledge, the first iPhone enabled Doc doing house calls. And I think I read about him in fast company or somewhere like that. And then later on, founded Sherpa that's now part of cross over health, real pioneer and virtual primary care. Now runs an organization called Automate.clinic.train AI models to think like doctors and tap their knowledge and all that. We're going to hear more about that. Of course, but he really brings a unique perspective of understanding the technical possibilities by, but really with a humanity and a human reality to how we get excellent healthcare. And it really connects directly to the keynote that I'm working on that's talking about this tension between, you know, as AI going to be sort of this tax and issue for our community to what could be hopefully. And I think it's in our control to make that happen. So Jay, welcome to re-localizing health. Before we dive into some of these topics, maybe just add a little color to what I provided in the intro in terms of so people can understand your journey and your words from that kind of early iPhone house call work to through Sherpa and where you're at now. Yeah, that's that. I'll try to distill 18 years or whatever down into the people that say. So yeah, I trained as a pediatrician and preventive medicine doc. And the iPhone came out the day before my last day of residency. And I've always been a dork and so really painted attention to like what the power in the internet was flicker was kind of my introduction to that, you know, that whole world. And I started a house call practice in Williamsburg, Brooklyn that was powered by my iPhone. And it went viral in this right place, right time. And sort of seven and a half million had some eyesight in the first month. And it was kind of like, you know, Marcus, well, I'll be old school doctor powered by the internet. It was something Vax news could kind of like bottle up and sell to my parents. So that was kind of fun. But then, you know, it wasn't like exactly what I wanted to do, but at the same time, didn't know exactly what I wanted to do because you can't really predict the future. So that led to a company called Hello Health and we built physical practices. They were membership based in places like, you know, the West Village or Waysburg, Brooklyn, but also Paul Combs employee health center, that sort of thing. So, and then, you know, after that started the company called Sherpa. And Sherpa was the pioneer as a virtual primary care. Prior to Sherpa, Hello Health was very much a, you know, phone or video transactions or 10 minutes with a stranger. And if you're a doctor, you know that like, well, 10 minutes with somebody I can never see it again, you're pretty limited with what you can do. So I said, well, hey, what if every time somebody picked up their app, you could work with the same doctor. And so you treat that more as like, you know, kind of no school concierge practice that was project managing your conditions over time. And so, you know, we found that 80% of things never really needed to be seen in person if the doctor could or labs or imaging, you know, prescribe, just follow up on your hand that direct relationship with you. So that was a pretty cool concept. Because that was, you know, it was, it was kind of early days of, I think, I don't know, just what healthcare was doing on the thing. Yeah, yeah. Well, you know, recently, you know, you've been writing about kind of breaking down healthcare into what I think you called that, Potomac Elements of Care. So when you look at a typical clinic visit, where do you see AI adding genuine value to that experience from both patient and the physician or clinician front? And what are the elements you think where human relationships are irreplaceable versus where AI can be super helpful? Yeah, it's complicated question, right? I think a lot of it is not so much about the technology but more about how do you human engage and help us into the workflow, you know? So, in a lot of ways, I mean, history gathering is, you know, a fascinating concept. The Sherpa, what we did was kind of build like 200 or about 300 templates for asking questions. So if you come at me with a cough and you're like a 40 year old, you know, female, what are those 20 magic questions that every doctor should ask, you know? And sort of, you know, that's a great way to gather history, but even more engaging, what is sort of have a chat about it. That's really trained on like those 20 best questions for every condition. And also, the ability to just kind of like go down the right rabbit hole. So it's not just 20 questions, maybe it's 28 and those rabbit holes were explored. So I can imagine a situation where you know, it's really, you know, easy for a condition to just pop in the room, you know, right before pop in the room, just kind of like read a summary of like what's been going on with you in the last like 24 hours, 40 hours. So they don't have to spend, you know, 60% of their time with you asking all those questions that could have been done beforehand. So that's just a simple version of, you know, where I could add a ton of value. You know, as I even go back to the the Sherpa days, I thought about, you know, there's obviously every person in a physician's panel or in, you know, any organization or grouping of people, they don't have the same need, right? Polychronic senior is quite different than, you know, young healthy person. And a lot of times the waiting rooms are clogged with people who don't need to be there. And the people who do need to get in there more urgently or sometimes, you know, it means not unusual right now for it to get to be 30 days to get into a primary care doctor. Like, is that even primary care? Even it takes that long to actually have real contact. But some of you you wrote about struck me. I think it was like, I think you said, if clinicians offload much of their work to AI and achieve 10x productivity, they could see more patients or they could do other things. But, you know, maintaining relationships with 200 people while providing excellent care. That's, that's not easy. You know, can you kind of unpack this tension and kind of wear, you know, the best of your ability to see where things are headed? What does after I like do for doctors? Well, you hear about these 10x engineers. What's the, does the concept of a 10x doctor exist? And if that does, right now they see about PCP, you see about 2000 patients in their panel. And like, if it's 20,000 patients, what does that do to relationships? I mean, it's like way past the Dumbart number. And so what can you do? Whenever the relationship is taken away and it's all about just efficiency going forward, does it seem like a world that patients and or doctors win? Well, I'm sure you have some hypotheses, right? The year testing and you think about just a general purpose AI, you know, periodically. I think I just saw some headline about Microsoft was claiming their tool was four times effective as, you know, some particular scenario. You know, good for headlines. But I don't for all the reasons you're talking about. and I don't think that's going to be just like suddenly, oh, okay, that's great. So as you sort of think about the delta between just a general purpose AI, whether it's clot or, you know, chat GPT, what do you think is different about medical decision making that needs that kind of gap, so. - Yeah, I mean, I think decision making is the operative word there. That's not really how about Hemsworth right now. - Yeah. - And so the decision making part is this kind of new paradigm that needs to be created over the next decade to figure out what, how does artificial intelligence then click a doctor, you know? And that's where I think we're stuck. So I think that it's going to be a lot of engineering approaches that are going to have a few years to basically, you know, kind of raised to hit that wall where like you're kind of exhausting on these engineering approaches. The world's data in many ways has been exhausted, you know? So a lot of these companies are commanding walls in terms of their accuracy. They still work like myjek though, you know? But at the same time, you know, whenever you see these things sort of struggle with the medical edge case and make things up, you know, is that ever going to stop? - No. - Seems like an engineering problem, other than like a human problem with this point. So yeah, I don't know. It's such an interesting tone trying to be alive and be a part of this industry because there's somebody unknowns, you know? - Yeah. Well, and obviously the stakes are super high, you know, when you're in a medical context, you know, from, you know, obviously the obvious of life and death, you know, and then on the negative, you know, malpractice. And one of the things I wonder about is, you know, given the social determinants, right? 80% to 90% of health outcomes driven by things outside the clinic. It makes me wonder if that's actually, you know, a place where, you know, beyond the kind of formal medical treatment, whether it's housing, income, food security, social connections purpose. And even just like, there's, you know, a scenario where in these kind of fully actualized primary care organizations, team-based model that are dealing with very complex populations I'm thinking about, like the Nuke model in South Central Alaska with Alaska Native peoples or the Iover Health, you know, that we're doing, you know, senior care. And my folks, you know, my dad's five-year Parkinson's journey was made much better by that and followed Roshika Fernanda Pula, who's touted that. And I look at that and, you know, a lot of times the most important specialist referral is it to a neurologist? It might be to a health coach or a social worker. - Yeah. - And, you know, and it might be, you know, they need to go to a community garden or a walking group or a grief support, you know, circle. And, you know, as you think about that, what do you think it would look like for an AI to become more sophisticated at identifying these kind of social prescriptions, which are pretty low risk, you know, 'cause it's, you know, whatever, socializing versus isolation. Like, I don't think anybody's gonna get sued for encouraging the people be more social, you know, rather than just medical ones. Have you thought of all about that part of it? - Yeah, I mean, I like to think of this as there's help in them, there's medical. Medical is pills and scalpals and doctors and, you know, appointments and insurance. Health is just like every day, habits, you know, and routines, relationships, the food, you eat, that type of thing. And so I find that fascinating. Because health has always been such a more engaging thing. And it's not patriarchal. It's more how, you know, how can we engage with the more, be you more content that you might keep out about? So in a lot of ways, like the health world is always ahead of the medical world. And so what is the health world doing right now with AI? You know, it's interesting to think about this, but I think it's just, you know, again, up in AI and others are just consuming of this health content. And so people are able to, you know, just chat back and forth and have every question and answer that you could ever want. It's engaging at the same time, like, you know, are people truly coming together in the real world? Because of these interactions, I really hope that somebody is able to pick up that out. Sort of like, you know, meet up with started in like, 2009, like what's, what would meet up look like if it were started AI first way? It's going to be 25. You know, I think that's pretty fascinating to think about. Yeah, yeah, definitely. There's a staff that kind of was shocking, which is we know the largest healthcare economy in the world is the US healthcare system, like we're spending the most dollars. And after that is Germany, but way after it, you know, it's like five trillion versus 500 billion, or over 500 billion, but not far behind Germany is the US out of pocket spending, right? That's literally the third largest healthcare economy in the world. And I think that, you know, and it's related, but separate from that, you know, 30, 50 billion dollar economy that's the direct specialty care and direct primary care and kind of the things that are happening in the health Rosetta land and other kind of related ecosystems. But I wonder if you thought about that in terms of, I mean, part of this is the, obviously, there's a lot of direct to consumer stuff on the edges, whether it's fertility or it's, you know, testosterone replacement or, you know, whatever, I mean, you see some fairly significant groups there. Do you think that's actually where we'll see the, the kind of innovation coming in kind of these, whether it's direct to employer, direct consumer, that that's actually the place where this is going to happen more than inside the traditional walls. Yeah, that's what's fun. Like a lot of people are talking about this sort of wealth transfer from boomers to, you know, generations, you know, after. And that's going to happen basically this decade, right? This next 10 years. And I always like to think of this as like, is there a health transfer? Is there a health experience transfer? You know, what would healthcare look like if it were designed by, you know, Gen. Like, right? Yeah. And that what is absolutely what I think is fascinating, you know, you look at things like general medicine right now, and I've seen them, but early at home, and DJ Parker from Pillpack, no, I'm sorry. I started a new company. That's trying to make, you know, these transactions that easily purchaseable. I think that's fascinating. So I think that these companies that speak to the younger generation are going to win at the same time, you know, someone yearned yearned me, you know, a week ago, had emergency, you know, surgery because of an internal hernia. She would have died, you know. I don't know where those services fit into that. But at the same time, when you look at Epic, you look at what, you know, my chart is doing for just communicating in the in-betweens with patients and sort of coordinating care and giving them the directions and setting expectations where it's, you know, it's refreshing to see them do a pretty good job, you know, at that. So I don't know. And I'm always looking for what I call like little utopias. And the little utopias are healthcare delivery that happens outside the sort of traditional peanut model. So that's where I've always operated. That's what I'm most excited about for sure. I think the other piece that's interesting is you now have, I think 70, 75% of workforce are millennials or Gen Z, right? Like it's not a little fringe thing. They will, this millennials are about mid 40s. (upbeat music) If you've been listening and thinking, I wish there were more places where people share what actually works. That's exactly why we created Rosetta Fest. Jonas and Nashville at the end of July, it's where employers, unions and clinicians who've built thousands of health plan successes share how they improve the caregiver and patient experience that leads to the best health and financial outcomes. The only people on stage at Rosetta Fest have created sustained success and happily share their secrets. Details and tickets at Rosetta Fest.org. All right, back to the episode. (upbeat music) And you know, we have a system that was really architected in the era where the greatest generation boomers were the primary focus in acute conditions. You know, today it's more about chronic conditions, mental health that are more preeminent. So it is interesting. to see how that that will all play out. But that's, you know, that's what I think is so fascinating about like AI right now is what is the role of conversations that happen in between. That's yeah, you know, how do these things happen? Why do they happen? They kind of have to be baked into like the operating system of like say Apple help or something like that, you know, but I could see that being massive like and then I just kind of like pin you proactively in a regular basis just to say hey, how's your day on? What's going on? What's going on with your health? You know, what's going on with the specific condition? And then you know, identifying when flies, alerting your providers that kind of, you know, care for you. That like, hey, something's weird, something's off with this person, you know, that's where AI is going to be just perfect. And you know, I always like to think of a healthcare journey is like kind of a novel and that visit with the doctors a paragraph within a chapter at that novel, but there's a whole novel being written. And so, you know, how does AI feel? The rest of the pages of that novel? Yeah, yeah, no, I think that's right on. And one of the stats I was sharing recently was that the average patient spends 16 times more time watching pharmaceutical ads and they spend with their doctor every year. That's a big part of the in-between time. You said we're going to be stats, man. Yeah, yeah. Yeah, in high school, I was my nickname was stats. I've read stock in my brain, but the other thing that I think is interesting, there's a company that we've done some work with called converging health. And they, you know, if you look at, there's kind of broadly three buckets of patients that they've identified. There's like low risk. There's really nothing going on. Just keep up with basic screenings and preventive care done a dot. And then there's the ones who are in the mix, big spenders, frequent travelers in the system and complex care, all that. Like they get appropriately a lot of attention. What they've done through their analytics and what not is there's this in-between bucket that are rising risk, they will move into that high cost bucket. And through their data, they can identify like 70% of those who will become the high risk. And I would think that they would be able to be that much more effective if there's more that in-between communication, right? The stuff that, you know, sometimes, you know, just as a doc, sleeping in the room, they're like, oh, one more thing, you know. And I'm even back early days, right? When PCs were coming into the mix way back when I was at Microsoft starting their healthcare deal. One of the things I remember was it was a kind of early in the HR space and they said, you know what? When people fill in the intake in a computer, it's kind of the early tablet computers, they're actually more candid than when they're actually talking to a doc. Yeah, I'm going to this time. Yeah. Which is pretty interesting too. And again, all those things, right? You can imagine how much more helpful AI could be in interpreting that. And then surfacing, okay, what's, you know, a certain point. It's not just not knowing that. What are you going to do with it? So, I was like, the server is 95% asynchronous messaging between our doc patients. And every single doctor was like, my history is 10 times better in this format than it is in normal conversation. You need to do absolutely talk and work openly. When they don't have to look, some kind of mostly, most of the families strange in their eyes. But something you said that it is interesting. There was an HBR called a few years ago put out by the forensic Kaiser. It was just called managing the most inexpensive patients and most complex patients. And basically they talked about that same thing. There's the 5% of people that are responsible for 50% of care, right? And they found that there's three personas within that 5% and the one that's like, you know, it's just, it's a car accident. They need surgery. You know, what do you do? Right? The other one is, they're going to die within a few months. What do you do? But then like the third component, and it's about a third or third. The third component is like that person with diabetes, who's, you know, not super savvy. And can you help them? And you know, they basically said, okay, that 5% is like a third that you can actually target. And it's that group, you know? And so it's really interesting to me because it probably is like all this effort into that, that, that, you know, basically what 1.8% of people. And found that thing, like there wasn't really a lot, you know? And what I think is so interesting about that is if you flip that around and you say, well, if 5% of people are responsible for 50% of care, 95% of people are responsible for 50% of care. And the 95% need really basic stuff, you know? And that's just pure primary care. And so if you actually impact 20% of that spend, you do way, you have much more of an impact focusing on just the basic stuff than you do focusing on that third of the 5% that you're like, you know, you may not get, you know, sort of an overall healing. So I think that's mostly again, that's kind of like why, you know, I'm super excited about like all these new services that are coming in, sort of sure, but like in a lot of ways because it's really focusing on that 95% that need just basic good access. Yeah, yeah, no, great. I mean, as a clinician yourself, what would be your advice to various clinicians, whether it's physicians, nurse practitioners, PT, whatever, unnavigating this transformation? Like, what would your advice beat it down in terms of how do they prepare for what should they be following, you know, while preserving what drew them into the field that, you know, they chose? Yeah, that's a hard one. You know, one of the, cash, one of the, of course, find this sort of, you know, folks online that are that are deep in the space and follow them. But at the same time, like I would recommend, and it's easy for me to say this, taking at a risk, taking a hit, and joining, you know, this new wave, this new future. It's just so easy to just kind of take the job at the house, one of the clinic in your neighborhood. It's what we know, it's what we do. But at the same time, you know, if you can just convince yourself to kind of go against that sort creativity that was being beaten out of you in my school to become like part of the future. And also, like, build those practices, you know, trying to figure out like, I think what's so interesting about these new services, if you sort of primary care practice here in Boulder, for example, like it's going to be full really soon. Just hang out a single, pick some insurance. If you start a, you know, really cutting edge online practice, like nobody knows you exist. And so it's just really hard to, let me just really track that in about how to acquire patients if you're not sitting in someone's neighborhood, you know, so that's just a big and important problem to try to solve. And so I would just put a lot of these, a lot of your like efforts into working for companies that are trying to solve the problem too. Maybe just last question, you know, for regular folks, if you're going to leave, how people leave with this conversation with one insight about AI's role in healthcare transformation, you just had to pick one thing. What do you think it should be? Tom Bann, that's a great question. I don't know. I would just say, be ridiculously curious, you know, like literally nobody knows what's going to happen here. And I think the most important thing is to just try all of these tools because if you say, well, you know, this is medicine, these tools are, you know, certified, qualified, like, you know, I know they're safe. Like, that's the type of attitude, you know, that we saw whenever one see a procedure was invented, you know, I don't know if you've ever heard that story. And a seizure was invented. The surgeons of the day were like, we don't want that stuff. We listen to the screams and moons of our patients to guide our surgery, you know. And so curiosity is that thing, a room that like I think changes healthcare. And so I know thinking differently as clinicians is not well respected in the in the industry. But that's the thing that takes us forward. So to me, it's just think creatively, think sure, be curious. And, you know, just ride the wave. So, you know, have fun. Really appreciate you. kind of helping us, you know, be candid about your journey. And, you know, I think that we're not here setting Pauliana, but we do think that AI, if child right, can be a huge liberator. And, you know, for folks who want to be a part of that transition, you know, I mentioned that we are of that Rosetta Fast. So, you know, we're really building that next economy. And with the the four booking folks on the clinical side, we can't do it without them. And those curious folks who are frankly fed up, you know, with the old system, getting together with the folks on the the payment side. So definitely encourage folks to follow Jay's work automate.clinic, because at the right URL. Yeah. And, you know, you can find more information on RosettaFest, just RosettaFest.org. But thanks again for everybody. And especially thanks to Dr. Jay Parkinson. And be a part of this whole transformation. And I'll continue to follow your work and I'm sure we'll stay in touch. So thanks again for joining us and thanks to everybody for listening to Relocalizing Health.