.png)
The Clinical Realist
Healthcare innovation is broken. We have billion-dollar AI running on 1990s infrastructure. We have startups dying in "Pilotitis." And we have leaders frozen by analysis paralysis.
Dr. Sarah Matt (The Clinical Realist) is here to fix the disconnect between the tech stack and the trauma bay.
Join Dr. Matt—physician, strategist, and author of The Borderless Healthcare Revolution—as she cuts through the hype to reveal what actually works in modern medicine. No buzzwords. No fluff. Just the raw, unvarnished truth about how to lead, build, and survive in the future of healthcare.
If you are tired of the "Star Trek" vision and want the "Clinical Reality," this is your show.
Subscribe to The Sarah Matt Briefing for weekly insights on healthcare AI, access strategy, and the business of medicine: https://drsarahmatt.com/newsletter-signup
The Clinical Realist
AI Is a Power Shift, Not Just a Tool: Dr. Robert Pearl on What Generative AI Actually Changes for Physicians and Health Systems
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
—
Resources & Links:
📖 Get the Book: "The Borderless Healthcare Revolution" is available now on Amazon and major retailers.
💼 Work with Dr. Matt:
Looking for a keynote speaker or strategic advisor?
Visit: drsarahmatt.com
🔗 Connect on Social:
LinkedIn: https://www.linkedin.com/in/sarahmattmd/
YouTube: https://www.youtube.com/@DrSarahMatt-ClinicalRealist
📧 Subscribe to The Briefing: drsarahmatt.com/newsletter-signup
—
Disclaimer:
The views expressed on this podcast are those of Dr. Sarah Matt and her guests. They do not necessarily reflect the official policy or position of any affiliated institutions. This content is for informational and educational purposes only and does not constitute medical advice or a professional consulting relationship.
Two years ago, every health system was running an AI pilot. And today they're still running AI pilots. So something has changed and something has not. The technology got dramatically better, but the deployment gap, the distance between what a board approves and what actually happens in the exam room, it's pretty pretty much the same. So today I'm talking with Dr. Robert Peral. He is the previous CEO of the Permanente Medical Group and co-CEO of Kaiser Permanente for 18 years. He ran one of the largest physician-led systems in the world, practicing surgery at the same time. And he's now at Stanford teaching the next generation of healthcare executives. And we're going to talk about his book, ChIGPTMD. So Robert's argument is simple and maybe a little uncomfortable to some of our listeners. AI isn't just a productivity tool, it's a power shift. And the way physicians and health systems are responding to it is going to determine whether patients benefit from it or get left out entirely. So let's get into it. So, Robert, thank you so much for being here. Before we get into the substance, I'd love 30 seconds on what's keeping you busy right now. What are you watching most closely in AI?
SPEAKER_01The main thing I'm watching is the move from AI becoming being mainly an administrative tool or a cost reduction tool to one that can save hundreds of thousands of lives, provide access to all Americans, and make healthcare more affordable. I'll start with just one statistic. According to the CDC, if we could effectively control chronic diseases, we're talking about hypertension, diabetes, we could eliminate 30 to 50% of heart attack, strokes, kidney failures, and cancers. Just have listeners imagine not just the improvements in health, but how much less healthcare would cost if we all avoided heart attack, strokes, kidney failures, and cancers.
SPEAKER_00I would love to avoid all of those. Absolutely. Well, you ran Kaiser through two very distinct eras of health AI. So give me kind of an honest before and after. What actually changed chronically between the first wave that narrow AI, rural space engines, et cetera, and what we're seeing now with more generative AI, agentic AI. And I don't want the press release. I want your real opinion.
SPEAKER_01Well, again, for the listeners, let's just start with those two uh polls because they're completely different. We should never just talk about AI. So narrow AI that you're referring to is when the computer application takes a data set, maybe 10,000 mammograms, 5,000 show cancer, 5,000 are either normal or benign disease, and then it finds 30, 40, 50 different slight variations of these two data sets between malignancy and non-malignancy. And with that, it assigns a probability factor. And now when you give it more mammograms, it's able to tell you with a high probability, in fact, about 10% better than doctors, whether it's cancer or not cancer. Generative AI is completely different. It's never, it's not given data sets. It's uh taken trillions of words from the internet, from online sources, from textbooks, from journals. It has all of the information that doctors learn in medical school and in residency training. And then with that, and about two billion parameters, it's then able to answer any question. It's able to personalize it to you. And I want to say the subtitle of my book, ChatGPTMD, is how AI Empowered Patients and Doctors Can Take Back Control of Medicine because the patient has to be first. And that's what is different. And I don't know anyone in the United States today who is using a genitive AI tool to empower the patient to be able to both manage their own chronic disease, to understand and diagnose their acute disease, but we also do know that about a third of Americans are doing that on their own. So we are at the dawn of a new era.
SPEAKER_00So when you think about boots on the ground physicians in the exam room, how do they experience medicine today differently than they did two years ago when we were in that narrow space? Or don't they?
SPEAKER_01Well, well, the biggest thing they have right now is ambient listening chart generation. So two years ago, they were staring at the computer screen, typing away, and today they're able to look at the patient and have the information available. That's a major advance. I don't want to in any way minimize that. But, well, there's a second way they're using it in the exam room today, which is that there's a couple of sources of information. One's called open evidence, one's called up to date, which allows them to consult. And about two-thirds of doctors today are consulting it in the exam room when they are stumped. Because remember, medical information is changing, it's doubling every 72 days. There's no physician who can keep up. When you see a complex patient who may have a rare disease, you need to have that information. You would have normally gone to the library, researched it for a couple of hours. Now you can get it directly on your computer. But there's just a study that was published out of NYU that showed that people who use a large language model are able to get as much expertise as someone who has one of these more expensive and curating opportunities. And so I think that physicians are using it for their own information to expand their expertise. What they're not yet doing is teaching patients, and I don't mean personally, they can give them a link, they can give them a video, they can give them handouts about how to not only prompt a generative AI tool, but then ask the appropriate follow-up questions. And if we did that, now we could change the exam room from being an episodic calendar based. You have, let's say, hypertension. I give you a medication, I say, come back in three or four months. And what happens in between? I have no idea. And the last thing I want is 100 blood pressure readings coming from you to my office. I just can't manage that. But what I would like to know is how are you doing? I'd like to have, let's just say, three readings a day. After a month, 100 readings, first and second derivative of the uh trend. You're getting better? I don't need to see you. I don't have to bring you back for a check in three or four months. But if you're not getting better, I'd like to alter your medication to be able to get better control. And we know that hypertension is the leading cause of strokes. 40% of strokes happen out of hypertension. Imagine if we could now control that at home with a electronic blood pressure cup. It's $25 at Costco, now connected into a with through Bluetooth into a general application. With that being able to let you know as the patient how are you doing, I think that would be a massive advance in the practice of medicine.
SPEAKER_00So it's interesting. In a previous episode, I was on with Amanda Brewitt, and we were talking all about patient experience. And we think about the hospitality aspects of healthcare. It's difficult because when margins are slim, it's the first thing to go. We actually talked about how we could have patients have proper prompting and then available resources as well. Um, I know when I was younger, my mother had a giant blue book. And if anyone had something going on, it was the blue book. I'm the first doctor in my family. So when I became a doctor, I was basically competing against the blue book. Um, these days, we joke that the other doctor, Matt, because again, I'm the only one except for my mother who thinks that she's just as smart with her Google and her agentic and all the other goodness. Um, but all of our patients are doing that. I know that when I see my charity patients or when people are in their offices, we consistently have to help them understand the resources that they have seen, whether it's a TikTok or, you know, up to date that someone printed out for them and tell them, hey, you probably don't have a brain tumor. But so when we think about empowering the patients and using all the tools at our disposal, I love the idea of giving them proper prompting so they can look into it better. What would be a way to do that from an economic perspective to make it worth the medical system's while to build it out? I've found that patient engagement is usually paid for by hospitals and provider practices. And it's really just for their own benefit. It's for them to gain information and discrete data when you really look at it. What do you think?
SPEAKER_01So let me start. And I really love that vision of your mom in the uh blue book. I know exactly the book you're talking about.
SPEAKER_00You know the book, right?
SPEAKER_01I know the book. But let's look at that or let's look at a Google search. What does that give you? It gives you general information, gives you information about every patient with hypertension. Doesn't tell you about yourself, doesn't include information on your diet, it doesn't include information around the other diseases you might have, doesn't have any information about your socioeconomic status, how much money you want to spend on food each week. It can't personalize that. To me, it's the difference between information and expertise. Now, generative AI gives you that. So we have to teach patients how do you prompt that? How do you put your medical information into the system without exposing your privacy? How do you word that? You don't want to put your name in place, you don't want to say it's about me. You'd like to be able to say, let's assume there's a you know, 35, 40-year-old woman with hypertension, newly diagnosed hypertension, and it's going to want to know if you ask the follow-up questions. Well, are you pregnant? Because hypertension in pregnancy, as you well know, is very different than hypertension not in pregnancy. It may very well want to know about other problems. It may want to ask you about taking blood pressure in both arms instead of one arm. There's a whole list of things that we would do, and the technology can allow you to do that. That's what you used to go to the doctor for at the start of the problem. Now you, as the patient, the empowered patient, can do that. It's going to be particularly relevant in rural places and in socioeconomically challenged neighborhoods where maybe it's hard to get into a clinician, but I think it's going to be great for everyone because I mean that's where I start today. Before I call a doctor about a problem that I might have, I always want to get my most expertise possible. That's a very different view. And as a doctor, you know, culturally, that's contrary to how we were trained and how the practice of medicine has been for decades, if not centuries.
SPEAKER_00So I know your book was called ChatGPTMD, and the MD is doing a lot of work in that title. When we think about patients using AI to almost practice medicine, and again, getting more resources for themselves, is that finally correcting an information asymmetry that physicians have created over the course of time? And who's benefiting from that?
SPEAKER_01It definitely is uh addressing that um inequality between the clinician and the patient. But I don't blame clinicians for it. I mean, how are patients going to know? I mean, sure there were books out there, but that's a lot of reading and you don't have the background. I mean, remember the first days of medical school and you start hearing words and you have no idea what they mean. And ultimately, yeah, you spend four hours and a hundred hours a week learning what those things uh might be. But you know, early on after the book was published, I was on a podcast. And the at the end of the podcast, we were done recording, the interviewer said to me, Let me ask you a question, Dr. Pearl. You're you're a skier, you're a uh clinician. My husband fell three months ago. His arm was over his head, he slid about 100 feet, and his shoulder still hurts, and that arm doesn't work as well as the opposite side. What's going on? And I said, I think I know what's going on, but why don't you do this? You said you'd never use a chat GPT for any any of the large language models. I keep talking about ChatGPT because I like GPTMD, but you know, it's also true of Claude, and it's just true of Gemini, and it's probably true of a lot of the free models that are out there right now. I said, Why don't you go there and put all the information into the system? See what it says. And then if you still have questions, call me back. Five days later, she calls me back. She says, Thank you so much. I put all the information in place and it said he probably has a rotator cuff tear. What's the rotator cuff? It explained to her what the rotator cuff was. It said he probably needs an MRI. Why does he need MRI? Well, it explained the fact that you could see the internal anatomy around it. And you should consult an orthopedic surgeon because he almost definitely needs a procedure done. Now compare that to the Google search, the information you would have gotten. And she said, I went to the doctor and he said you probably have a rotator cuff, Terry. I knew what the rotator cuff was. I didn't have to ask. I could now learn more about it. I need to get an MRI. I didn't have to question whether he was doing it because he wanted to upsell me. I knew I needed to get it done. And at the end of the procedure, he said, if you'd waited three more months, I probably could not have taken the tendon and reattached to the bone because the muscle would have shrunk. That is an empowered patient. I don't like to think about it as a competition with the doctor. Now you have a true patient partner. And I'm sure as a clinician, you would agree you want to have the patient be your partner, not be undached from the healthcare problems they might be having.
SPEAKER_00Of course. An engaged patient is always much more reasonable to have that partnership with. You know, I'm very much on the positive side of technology and AI as tools with us to use with the appropriate use cases and controls. Um, but I know a lot of folks, especially in straight-up medical delivery, are just very frightened of it. And a lot of physicians in particular are concerned that the AI second opinion is gonna be full of liability and hallucinations. But I keep looking at those same studies you're looking at around AI error rates and physician diagnostic error rates. And the AI is getting really good. And Dr. Matt gets tired and Dr. Matt gets hangry. But you know who doesn't? AI. So when you think about your most, I'd say, disillusioned and skeptical providers, what are your thoughts on how to help them come along on the ride and give them that peace of mind?
SPEAKER_01Well, the first thing I say to everyone, whether they are a provider of care or a recipient of care, and of course many of us are both, uh, is try it out. If you're a uh patient, access the information and see what your doctor says to you. You're gonna be surprised 98% of the time, based on a lot of the data, it's gonna be exactly the same between the two. And as the provider, pretend you're the patient. Put the information in place, see what it would say. As I say, I fortunately I won't violate hypocrites about me. I'm in very good health. I don't have any chronic disease, I don't have any other problems, but I'm a runner. I run a lot. And I have a lot of running injuries, and running injuries are annoying. And as runners, you don't want to stop running. So before, but before I call the whether it's the podiatrist or the orthopedic surgeon, um, I check with Chat CPT. I want to know as much information as I can. And I can tell you that at least over the past 12 months, there's been no difference of opinion between the two. Almost identical, or at least as similar as if I saw two different doctors, because every clinician has a slightly different take on things, but overall the message is exact. So, first thing I would say is actually try it. But I'm now gonna say a second part. And this I say more to the industry than to any individual. The time has come for us in medicine to recognize the challenges that we have in the care that we provide. 400,000 people die annually from misdiagnoses. Another quarter of a million die from preventable medical errors. We control chronic disease, hypertension, we control 50% of the time. When I was the CEO in Kaiser Permanente, we controlled it 90%. How did we do it? We had a lot of nurses, a lot of nurses who were calling patients all the time, checking their blood pressure. It can be done. We have the medications, we have the tools. Why doesn't everyone do it? It's just too expensive. Why did we do it in Kaiser Permanente? Because we were capitated, we were prepaid. We did best when the patient stayed healthy. Rest of the world, no, it's fee for service, and that is not the outcome that is there. But that doesn't mean it couldn't get done. And as soon as we start to say there may be as many as a million lives lost every year that we could prevent, could help, could keep someone alive, now we start to see this tool differently. You know, there was just a really interesting uh breakthrough with a uh AI generative AI tool, actually from OpenAI, that looked at what's called the Urdos problem. This is a mathematical problem that had been around for 80 years. It asks, how many dots can you put on a piece of paper all equidistant from itself? And Urdos 80 years ago said most likely it was a very um geometric design, horizontal lines, vertical lines sitting in play. And he challenged people, he offered a financial prize to anyone who could prove his theorem, or his hypothesis at least. Uh, and no one could. And now this problem is given to a generative AI tool by OpenAI, and he came up with a solution. And it did it by combining geometric expertise with algebraic numeric number theory, something that would never happen. Because you have specialists in both areas, but they don't communicate with each other. That's medicine today. And I think that we can now learn from that. But if you look at the response the mathematicians had, they saw it through their own life. This was going to diminish status, this was gonna create problems for them, they're gonna lose some control, and that is true. And I think that's what we're going through in medicine today, the same type of reaction to it. It's a normal reaction that we all have. But what we fail to see, what I encourage people to do, is to start with the patient and the problems. And when you start with the patient and the problems, you can start to see why a genetic AI tool connected to wearable monitors and home monitors and patient input and a whole variety of other ways, could improve outcomes for the patient. And then as a result of that, we change the economic model of medicine, be able to award clinicians equally well.
SPEAKER_00I love what you said there because in my career, even though I'm a surgeon by training, I've really spent most of my time on the product development strategy side in the tech sector. And so what I see the most is the tech's the easy part. It really is. Like I have zero concern about our technical ability to create X, Y, or Z. Easy. However, getting it adopted, commercializing it, making sure it actually is solving an urgent pervasive problem, that's hard. And the first thing to go from a product perspective is actually talking to your customers. So what we end up having is amazing solutions that people are trying to shove into problems as opposed to defining those pro problems right off the bat and saying, how can I solve this problem? Or what are the best ways to do it? And again, all of our agentich toolbox may be a great way to do that. But you know, everyone complains about the zombie fax machine, but it freaking works for other stuff, even though it's not particularly exciting, right?
SPEAKER_01Yeah. You're absolutely right. You know, I teach both the Stanford Medical School and the Stanford Business School. And the business school, so many of my students want to be entrepreneurs. And I tell them the truth. 90% of entrepreneurial companies fail. And I'm talking about 90% of great companies. They have great ideas. You define it exactly right because technology alone doesn't do a whole lot. You know, sort of reminiscent of the days where you put stick it notes on a computer. It doesn't work unless people are are are willing to change the flow, how they do work. As long as you're going to keep seeing people in your office with hypertension every three to four months, nothing's going to change. As soon as you start to say to the patient, this is the technology I want you to use. And I want you to let me know with the, you know, with the when the generative AI tool tells you that your hypertension is not improving, particularly if it's getting a little bit worse, so I can make a medication adjustment. But if not, I don't need to see you. Just let me know you're okay. You now have a totally different workflow. You've created a lot of time. Maybe a 30 of patients don't need to come in. And a third of them can be managed with a simple email or a video visit or a very short way to change the medication. But now you have the time for the third who really needs your expertise because their problems are very complex. And they don't have just hypertension, they also have diabetes, they also have uh psychological, you know, a little bit of depression. They have a whole series of things to say, nothing about the um environmental factors sitting around them. So that's where I think the opportunity goes. But as you say, you said so well, you've got to redefine that workflow. And that's difficult and challenging for anyone.
SPEAKER_00When I look at the American healthcare system, I of course would love to just burn it down and start from scratch, but we have patients who need care right now and yesterday. So we can't do that. So I'm very pragmatic in my approach because I think we have to just utilize the existing system as constraints and therefore build with it. Two of the things I see really that are here are our rural health access issues and our aging population. And when you put them both together, these are the folks with the most chronic diseases. These are the folks that have the hardest time accessing care in lots of ways. With all of the, I'd say, sexiness of rural health right now, all the funding that's going towards it, I admit I have not seen great examples where that new funding has actually produced amazing outcomes. What do you see in this area? Am I missing something?
SPEAKER_01My approach to problems is to really try to understand what's the fundamental challenge that's there. And I would separate the two groups that you provided. You said the rural areas and the uh more senior uh members of society, because I think they're they have the same challenging outcome, but the underlying driver is different. So, what happens in rural areas? The population is just too low. By definition, that's what a rural area is. And when you have a low population, you first of all, you have a lot more driving because things are spread farther apart. And number two, you just don't have enough people that are there. So I think they have that fundamental challenge. What's the problem in the senior citizens? They got too many diseases. They have three and four different chronic diseases, and the model that we have today doesn't work. I think generative AI is going to be a solution for both of them, but it'll be applied differently. Yes. So if you ask me about the rural area first, what I would say in that arena is it can't be solved because you're not going to get enough clinicians to go there. There's not enough patients to uh generate the revenue. And so you're going to need to use a tool to expand that network to be able to provide a lot of care with virtual visits, telemedicine visits. When you deal with the patient with multiple chronic diseases, the answer is they just need a lot more care than we're able to afford to provide to them today under the reimbursement system that's there. So my solution overall is we have to change the incentives. We have to move to some form of capitation. When you move to capitation, it doesn't solve the problem, by the way, but it starts to align the incentives. So now, what's the first incentive? You can't capitate a single doctor. I mean, that that's what some of the MIPS and the macro. No, it doesn't work. No. But we capitate a group of clinicians, and that, by the way, is the biggest hurdle because as clinicians joining together and being willing to accept the leadership structure goes against the culture that we were trained in in medical school and in residency. But as soon as you do that, you start to be clear in the rural area. We're not going to solve this problem except by bringing in technology that's now going to allow us at very low cost, whether it's with nurses, uh PAs, uh MDs, it doesn't matter. We have to figure all of that out. And the answer is going to be a lot of that is actually going to be technological. Because, you know, as you well know, as a technical expert, the first copy of the application you create is very expensive. Everything after that is almost free. It's all free after that. So we can reaction applications to provide uh to people in rural areas uh maybe 30% of the care. They don't need to see a clinician. The technology can do it, assuming that the technology is able to acknowledge when it can't do it, and then that's when you need to bring in assistance. But you don't necessarily have to bring in the one doctor in the community. We have clinicians across the United States. We can do that in ways that make more sense. We're just not going to do that in a FIFA system world because as soon as we do that, we lose income. And I say in ChatGPTMD, there are two things that will never be implemented in technology. Anything that slows doctors down, or at least they'll never embrace it. Or number two, anything that cuts into their income. And then it's not because doctors are a problem. That's everyone in this world does not want to have more roadblocks or less uh income coming in. But as soon as you have capitation, it accomplishes that. When you get to the um, I'll call it the more senior citizens, the big challenge, the big opportunity is going to be controlling their chronic diseases. And we do a terrible job of that today. So to the extent that we're now able to benefit, you know, think about that. How much more money does a clinician get? Not the same clinician, but the medical system get when they reverse obstructed carrier artery to the heart and unblock them in a way that prevents the heart attack versus preventing it in the first place. Far more expensive, $100,000, $200,000 for something that we could probably have done with a combination of a $5 a month statin and uh uh an improved diet and exercise program. Now, people will say, the second the patient, I tell the patients to do that, they don't do it. Well, yeah, but we don't give them the tools. We don't give them the way to be able to put that information into a genitive AI application. You know, every Sunday night my wife and I sit down and we say, what do you want to eat this weekend? We say, well, you know, we went out quite a, you know, a couple of events this this uh Saturday night. We ate a little bit too much. Let's cut back on our calories. We put in the kind of food we want to have, you know, what ethnicity in food. We put in information about anything about health. Again, we don't have it, but if we needed a low salt or a low uh uh cholesterol type of diet, we could put that in. And we can put in a price point. And then we not only get shopping lists, we get recipes for the entire week to accomplish that. We've now given the patient something they can use. My belief is patients want to get better. We just have made it difficult. We've just given them instructions without the tools. You know, we need to give them the catalysts, if you want to think from a chemical standpoint, to make these things happen more easily. Again, I see generative I tool being able to accomplish that.
SPEAKER_00So after a call here, I'm literally driving to a board meeting for a small healthcare system. And when we think about capitalisation, of course it makes sense more predictive methods, prevention methods. Now we're really eating our own dog food. But how, even at the board level, could we even consider those kinds of changes from the system we're using now? What do you think would be the best ways for hospital boards or other leadership within healthcare to start to make that change? We don't need more medicine macro, we don't need more meaningful use, we don't need anything like that to push us. How can we, as business professionals running these systems, make it happen?
SPEAKER_01Well, it's very hard. You know, it's like a relationship. It's very hard for one person to adjust the relationship. You need two people. So the first thing I would say is is there a willing um insurer? Is there some a payer in some way, maybe um a self-funded business? Is there someone in your neighborhood willing to partner with you? And if the answer is no, the answer is you can't do it. But if you can find someone who's willing to do that, uh, you need to create a financial relationship because the, as you well know, if you make all these changes to chronic disease, to hypertension and diabetes, you don't see the reduction in heart attacks, strokes, kidney failures, cancers for about five years. Yep. Sometimes 10. But there's a lag. And you've got to be able, as a health system, as providers of care, you've got to pay your nurses, you've got to pay your staff, you've got to pay your electricity. You need to be able to be able to manage across that. Uh the the best I can come up with with a model, and no one, I can't find anyone who's yet done it, but I'll encourage you, you can tell me, is to say, okay, for the next five years, we want you to pay us the average of what you pay everyone else. Whatever that is. Uh, let's assume that uh, you know, for the mix of patience, we'll we'll do all the usual kind of uh evaluation to make sure that it that it makes some sense. And that's what you're gonna pay us. We're not gonna send you any bills, and by the way, you just save 10% of your revenue because we spent 10% of revenue on uh billing and clothing. Yes, yeah, yeah. And we're gonna do a comparison of our cost against that average. And five years from now, we're gonna split it with you. So if if if we spend more than the average as a result of this change, then we'll pay half. But if we spend less, then you're gonna give us half. And I can't imagine that we will not drop that cost dramatically. We just start with the fact we eliminated billing and claims, uh, to be able to now share in that. You know, that that's in some way, and I can I could convince my physicians though, again, every organization's different, that this is the kind of thing you do when you go with a private equity or someone else. But we're not gonna go with someone else's private equity. We're gonna do it ourselves. And yeah, we're taking a risk. I want to make sure everyone's clear about that. But I have a, I know you, I know how you practice. I have a 90, 95, 98% confidence we're gonna win. And that's a good bet. And now you've got to get that at the board level and not just at the board level, the leadership level, and most importantly, you've got to get it across your entire organization. So I don't know how big they are, how many people they are, but this is what I would do. It's what I did when I was the CEO in Casa Permanente. You know, when I took the job, we had two days of cash on hand. You need three days to be able to meet the insurance requirements. We were in trouble. That's how I got selected. They had fired the previous CEO and they hired me to do it. And by the end of that particular process, by approaching it from the standpoint of how do we become not a sick care organization, but a healthcare organization. Give great sick care, don't get me wrong. But how do we actually focus on those opportunities that were there? And that's when we brought in technology. At that time was the electronic health record. It's when we brought in chronic disease management programs, it's when we did a lot of the things that were going to be there. You know, we got to the point where we had what we should have 60 days of cash on hand.
SPEAKER_00I love it. Well, we're getting to the close, so I want to make sure we have time for the lightning round. So a couple quick questions. So, as you're teaching med students today, what is the number one thing you think that med students need to understand in today's healthcare environment?
SPEAKER_01I think med students need to look at the future. We have to understand this is going to be now a revolution in academia. When med students arrived in the past, when you and I were in medical school, every one of our professors knew more than we did. Nine more. Now today that's what? Not anymore. It's exactly right. We don't have to teach them how to use the generative AI tool. They've already been using it for you know quite a number of years, and the future generations will even be more so. But they need to be thinking about how can this tool be applied when they finish their training. Remember, it's 10 years, as you know, it's 10 years from now. By the time you do four years of medical school and five or six years of uh residency in a lot of specialties, general surgery is seven years now, so it's 11 years. Uh, how are you going to use it? You know, think about this. You know, you do a procedure, you send the patient home, so an outpatient procedure, and you want to know if the wound's infected. What do we do? Well, if it's red, if you have a fever and pain happens, well, that's really subjective. How about a video camera looking at that wound every single day, evaluating it, asking the patient some questions? I think we could get a lot more accuracy so that you can see the people who have an infection three days later. We say, come back in a week. What's a week? You know, some infections are three days. How do we advance? I want you thinking medical students about how you're going to use this technology when you're finished. I'll warn you, your professors may not like it if you raise it now. So be a little quiet about it to protect yourself, but be thinking about that so that when you finish and you start your practice, you can organize it in the right ways. You can work with colleagues, you can create integration, you can do all these things. Yeah, the world's gonna be different. I teach strategy at the business school. Strategy is about the future. Maximizing your probability to maximize the probability of being a great doctor 10 years from now, you better figure out how just.