The Future of Healthcare and AI for Insurance with Doctor Mike

Doctor Mike and Sean Merat on Stage at Insurtech Insights

At Insurtech Insights 2026 in New York City, Owl.co Sean Merat interviewed board-certified Dr. Mikhail Varshavski, a.k.a. Doctor Mike, about the future of healthcare and AI for insurance.

Below is an executive summary of their discussion followed by a full transcript.

Summary

From his perspective as both a board-certified physician and a YouTube influencer with over 30-million followers across social media, Doctor Mike's advice for insurers with AI for healthcare is simple but important: prioritize a people-centric, patient-first approach to AI.

With a human-AI partnership, technology can eliminate administrative friction and support claims teams without replacing human judgment, ensuring AI builds trust rather than creating barriers to patient care.

The Challenges

Healthcare is hurting. Patients face friction that delays care while doctors burn out from endless paperwork and flawed documentation. Without institutional changes, including from insurers, patients will be further squeezed out and human connection strips away.

How AI Can Help

AI can step in. By handling paperwork and fixing documentation errors, it lifts the administrative burden off doctors. And by doing likewise for claims adjusters, who become liberated from documentation, they’ll have more time for what matters most: helping claimants.

Successful & Ethical AI Adoption

To make this work, insurers must put patients first. This means cultivating a human-AI partnership where AI only recommends, never decides. AI can handle low-stakes tasks, but humans must take over when medical risks are high.

Done right, AI builds trust instead of barriers, enabling claims adjusters to better support patients and empowering insurance executives to become industry stewards.

Transcript

Sean Merat:

Well, we're all super pumped to have you. You've built one of the largest doctor audiences on the planet by being a bridge between the public and the medical community to rebuild trust. And trust in the healthcare system has collapsed from 70% to like 40% since 2020 to 2024. What really broke, and who in this room can actually fix it?

Doctor Mike:

Yeah, I think a lot of things broke. Being that I create content on social media and I find myself in this unique intersection of marketing, social media, television, and my day-to-day job working at a community health center, and I see the struggles that my patients continue to have with the tremendous bureaucratic hoops they have to jump through, the consistent denials and fight that they have to have in order to get their healthcare expenses covered, some of my patients facing bankruptcy because of the bills that they face, they don't believe the healthcare system is working for them, with them. In fact, they view it as a complete adversary to good health. The pandemic certainly didn't do anyone any favors. The government didn't do a great job in handling that, both from an administrative front and a communication front. We made promises and sounded ultra confident when we should have shown humility and bridged our answers with uncertainty, the fact that we don't have all the answers. We should have been more confident in saying "we don't know."

And I think people just want the healthcare system to work for them more often. They don't want to feel that when they're going to see their doctor, that the primary visit, the primary purpose of that visit, is making the documentation of that visit, is that the doctor should make eye contact with them, not with their computer, that the doctor should recommend them care based on their expert opinion, and that's the care they should receive, not a middleman deciding whether or not that care is appropriate, especially when that middleman is not part of their care team, of that specialty. And that transparency has been gone for a really long time.

Sean Merat:

Your motto is "alert but not anxious." If you could get every insurance exec in this room to fund one preventative intervention, what would move the needle most, you would say, for the patient outcomes?

Doctor Mike:

I think that everyone in the room wants to do well. You want to make sure that patients are getting their care, and you want to do your job well. But we really have to figure out these incentives that are currently not aligning correctly. The incentives need to be so that if you help your patients get good quality care, if you prevent fraud without hurting patients, that's when you should be rewarded. And right now that's not happening. What's happening is we're rewarded for creating friction. And when we create friction, patients suffer. They don't have access to their healthcare plans, they find it difficult to see specialists, they are going into an emergency visit to an out-of-network hospital, not by choice, and leaving with a huge bill behind them.

So I think there needs to be a change of incentives in order to make sure that when we're creating policy, when we're working with AI companies, that we're always putting the patient first, and then that leads to financial success down the line. In fact, that's the primary motto by which we run our company. We are an advertising company in so many ways, and many people felt like we would have to take sponsorships that were questionable, that hurt our medical ethics, and we've proved that to be quite the opposite. The more selective we've become, the more that we demand companies that work with us tell the truth and that they're transparent, the more financial success we have, and the better outcomes for the campaigns that they institute, they have.

Sean Merat:

It sounds so simple, right?

Doctor Mike:

We are all primary care doctors, so I like to be practical.

Sean Merat:

I think it sounds so simple. In our company too, we, you know, we serve carriers but we're always thinking claimant-first. And I agree with you. I think when you think about the patient, the claimants, at the end of the [day], everything else falls into place, right? Now, you've talked about millennials relying on urgent care because primary care is inaccessible. What does this misrouting do really to the system downstream, and how should payers be thinking about this?

Doctor Mike:

Yeah, I think it's not just urgent-cares, I don't want to throw them under the bus, on their own. This is just a misapplication of technology, of availability, private equity getting involved in healthcare in ways that [are] not sustainable. When we create a technology, telemedicine as an example, this could be a great avenue for patients that have barriers to care. They can't leave their homes, they need to see a specialist that's incredibly far away, they don't need to actually go see a physician and take a day off work, because the conversation can be had by phone, because it's a conversation about blood-test results or changes in medication regimens. So there are uses for telemedicine. But then when we try to maximize and hyper-optimize the system to send all patients through telemedicine, we actually do a disservice. We get lower quality of care, because I've tried to do an abdominal exam via telemedicine and it does not go well. There is no world where I can be as accurate a diagnostician as I can be in real life [...] on telemedicine.

But the whole reality of being a good advocate for your patients is making sure that what issue they have is being appropriately managed. So if you're going in to manage your diabetes at an urgent care with a provider [who] has never seen you before, [who] doesn't know the challenges you have, you're setting them up for failure by making it seem like they're gaslighting you as a patient because they don't have enough time to learn who you are as an individual, what barriers you face, what insurance struggles you have, because they have 15 minutes and they need to move on.

So instead of creating more perceived gaslighting in the system, let's find the unique ways to treat patients appropriately in a way where we can reduce that friction and actually improve outcomes. There's actually a really good through-line here with healthcare in general. These days, if you turn on any podcast, it's all about hyper-optimization. Here's how you can run faster, here's how you can live forever. None of that is based on reality. A lot of it is based on preliminary research that is not yet true, that hasn't been proven true. It gets people excited, but it's not what actually leads to good outcomes. That hyper-optimization is a problem, because anytime in healthcare you hyper-optimize, you also create the opposite impact. Any medication, any surgery, any treatment always has side-effects. So if you're trying to hyper-optimize, there's always going to be negative downstream effects of that within our bodies. Same holds true for the healthcare system. If we try to hyper-optimize by trying to be overly strict with claims, with insurance coverages, we can actually squeeze patients out of their coverage and create more harm.

Sean Merat:

Now I'm curious, how did you try doing abdominal exam on telehealth?

Doctor Mike:

Asking the patient to perform it and give me their feedback. So, press on certain areas, let me know how that feels. And it's just, you're not going to train a patient to become a physician with 12 years of experience.

Sean Merat:

That's true. Now, you've, I think, talked a bit about AI in your practice. Where has AI earned your genuine trust in your practice, and where do you think it should be avoided?

Doctor Mike:

I don't think it's earned my full trust in any aspect yet, because we need to be alert, not anxious when it comes to AI as well. We need to make sure patient privacy comes first, patient protection always comes first in lieu of technical innovation at times. And that's okay, because just like our criminal-justice system, we are more comfortable with letting someone guilty go free than putting someone who is innocent into jail, we need to lead with that same approach in healthcare. Less patients should be hurt at the cost of innovation. It might be slower innovation, but we need to make sure that we [do] not leave patients behind.

I would love to see more AI innovation in the burden, the administrative burden that we feel as not just primary care doctors, but doctors in general. It takes multiple hours of administrative care to care for a patient for one hour in real life. And to me, that is a problem because the reality of what we signed up for as healthcare professionals is to be with the patients, is to deliver care, to make executive decisions, and not focus on secretarial work, checkboxes, making sure that we're meeting all these standards that are not actually improving care for our patients.

Sean Merat:

Yeah, you know, I think I actually saw you once talk about how the admin burden of what you're doing just completely sucks up all the time. And so on this side of the world, I think a lot of people in this room will resonate. So we help carriers with a very similar problem statement, right? They, the claims specialists, have to review thousands of pages of documents, some of them are medical, some of them are legal, some of them are financial, and their expectation of them is to review it all in 15 minutes to make decisions. And of course, I've tried, it takes hours and I make all the mistakes in the world, so I have all the empathy. So the idea is like where could we deploy AI where it would actually have a meaningful impact, take the mundane admin work out so that they could provide their human expertise where it really matters. Back in the day, there were buildings of people doing calculations of, like, accounting, you know, and now you just got an Excel sheet.

Doctor Mike:

Yeah, I think we're living 30 years behind the times in healthcare. Whether that's talking about faxes being sent, or pagers still being the primary mode of communication, I think we need to be more open to adapting technologies, especially when it's a low-risk probability. Like, this is where we can squeeze the most juice out of healthcare. Even for myself as a business owner, to try [to] figure out what insurance plan I want for myself and my employees is one of the most difficult tasks, and I'm someone with 12 years of experience; I've interviewed the head administrator of CMS on my podcast, and she has struggles doing the same thing. So clearly the system isn't working in a valuable way. We need better UI, we need better explanations, and we need to move away from this era of celebrating that if we create friction, if we create complexity, we can be more profitable. Let's think about how we can be more profitable by being more effective rather than less.

Sean Merat:

Right, I love that. But let's, in ideal world, let's say you had technology take all that mundane admin work off, what would that world look like in your world?

Doctor Mike:

Not only would I be able to see more patients, I would be able to see those patients happier myself. My colleagues would experience less moral injury, less burnout. Patients would feel less gaslighting. In fact, if you look at patient-reported surveys, gaslighting has been one of the most reported issues for patients over the last 10–20 years. But it's not that suddenly doctors have become cold, not caring. It's that the system has changed, the incentives have changed, that it's less valuable for a doctor to spend time with a patient getting to know them and more about the relative value units, how many checkboxes did they check, did they select the correct CPT code for the visit. And that's not where healthcare should be focused. It should be focused more on the doctor-patient side of things. I think the more that we can make things human, the better outcomes we'll get for doctors feeling happier, staying longer in the field, and patients feeling like they've been seen.

Sean Merat:

Yeah. So this, you know, with all of this admin burden, how complete do you think the notes and the charts that the physicians are making today? And how much do you think, down the line, claims decisions are being affected by incorrect or incomplete documentation?

Doctor Mike:

Yeah, I think one of the biggest issues is if I write a chart and it's incomplete, or if I follow up on a specialist chart and it's incomplete, it makes it difficult to deliver care, and at the same time, it's going to create issues down the line to make sure that their care is covered, that the medication that traditionally is second-line and should be instituted won't be because they didn't see the documentation earlier on that they tried the first-line medication and had side-effects with it. So I think having completeness, having AI create reminders for us that are not creating more administrative burdens is very valuable.

I mean, this whole AI spectrum really hits home for me as a doctor because I think of it as homeostasis. Homeostasis is the body maintaining a very strict balance. Not too hot, not too cold. Not too acidic, not too basic. In fact, the difference between a healthy acid-base level and an unhealthy acid-base level is like [in the] 0.3–0.4-range of a pH. So the human body needs this balance, and I think our healthcare system needs a better homeostasis when it comes to AI, technologies working together, and the insurers that oftentimes act as a barrier as opposed to someone who is supporting their patients.

Sean Merat:

You've warned [about] AI hallucinations and how [they] make up sources. Now, for an industry like insurance that's putting AI into claims, what's your test for [...] AI systems that you would trust? Like, what is that test [for] you?

Doctor Mike:

I think it needs to be the human-AI partnership. It needs to have human oversight. Too many times I'll read an answer that AI gives me, even if I'm using the highest-level model, it's either incomplete, missing nuance, [or] not understanding the human part of that equation. And that's because there is no human on the other side of that accepting that information. So if we have our best claims adjusters, top 10% of them that are doing well, get those people to continue working with AI and not use the people [who] are traditionally failing in the system, so we can actually maximize the benefit of the top 10% of claims adjusters and make the AI work more effectively.

Sean Merat:

Interesting. Yeah, so I would say, like, for us, there's just so much incentive, right? On the industry to deploy AI, like, to rapidly make decisions and everything else. And as an AI company, since before AI was cool, we think about, [well], we want to do the boring work, right? And we have to be ambassadors of "let humans make the decisions," "let AI, like, pull up some, surface some important information, make recommendations, but never make the decision." But we also think about, you know, claims adjusters, if, like, one of the challenges is, there's not that many claims adjusters, right? And there's very few, like, top-percentile claims examiners, and, a lot of, like, newer, you know, less experienced ones. So I think the technology could really be helpful [to] reduce that delta and really upskill.

Doctor Mike:

Yeah, I think there's definitely an opportunity there. What I want these systems to not miss out on is if they're making decisions based on the information that's provided to them, I don't want them to just make a decision or even make a recommendation. I want them to understand the consequences of the decision that it's making. So in the past, I've seen my patients get denied care for a medication, let's say, that would prevent their arthritis from destroying their joints to a permanent stage where you can no longer get that joint to be healthy again.

Sean Merat:

Yeah.

Doctor Mike:

So the AI needs to realize that if it's making a decision, what are the consequences of making that patient wait? Is there a risk? And if that risk is high, I think AI's role should be instantly lower value, as opposed to having a human reviewer in that case, because this could be the meaning of life and death, high quality of life, low quality of life, and right now that's missing in the equation. It's easy to say, "oh, let's have the AI approve or disapprove this medicine that is a long-term medicine that has substitutions, lower risk on that one." And in general, how long it takes without AI to get an answer changed. It's okay if AI makes a mistake in an issue like this. But when it's life or death or quality of life, that's where it needs to really do better.

Sean Merat:

I agree 100%. So, an honest moment I guess with all the insurers in the room: what do you think insurers do today, like the one thing that you would pick that is actually harming the patients that you see every day, and what is anything they could do in order to enhance, like move the needle?

Doctor Mike:

Yeah, this is a bit of harsh reality, but I think the insurance industry does hide behind fraud quite often. And I'm a business owner, and I understand fraud happens and it needs to be held in check, otherwise the system can break and then no one gets insurance. But we cannot use the notion of fraud to deny care where it's appropriate, or not even deny care, make the process [of] getting care so complicated, so many hoops to jump through, that ultimately people will just give up and take that first decision at face value and give up on getting quality care.

I think we need to improve the prior-authorization system, we need to revolutionize it. If we're going to do peer-review, this should be peers [who] are of our specialty who understand how to take care of these patients. I should not be giving education to a sub-subspecialist about how primary care works when I'm trying to advocate on behalf of my patients. So I think this whole system needs to be revolutionized and not just hide behind the fact that we're preventing fraud.

Sean Merat: Yeah, I agree 100% and, I see, I could talk to you for hours, but lights are going off, they want to kick us off the stage. Really appreciate the time that you've taken [to] come here. Really enjoyed the talk and thanks again.

Doctor Mike:

Thank you so much everybody!