Healthcare 2026: AI Doctors, GLP-1s, and Insurance Defection

If you go on the like our health insurance subreddit, extremely common question now of just hey, should I just not get health insurance at all? These are my premiums, this is kind of what I'm paying, this is kind of the drug I need or blah blah, blah. And for a lot of people, that answer is actually probably no.

0:00

The health care system that we have today in the US is drastically supply constrained. 23 million Americans work in healthcare, yet we've got long wait times. 100 million people plus don't have access to a primary care doc. 40 day plus wait times to see a doc if you can.

0:16

The existing health care system, they really want to create standardized care for people, right? And standardized guidelines to make medicine good at the median level for everybody, right? But at the same time, people want more agency in their health care in some capacity. No one wants to be told like hey, just wait and see.

0:33

What happens when Americans start rebuilding healthcare outside the traditional system. Not because they reject healthcare, but because the way care is accessed, priced and delivered no longer matches how people actually use it. For years, the US Healthcare system has been organized around insurance. The Affordable Care act expanded coverage and brought the uninsured rate down. And for a time, the system appeared to be stabilizing. But rising premiums, high deductibles and limited access have left many consumers paying more while using care less. In response, behavior is changing. Some people are opting out of insurance entirely. Others keep coverage, but increasingly pay out of pocket for diagnostics, preventative care memberships and digital tools that offer speed, clarity and control. Care is moving towards proactive screening, monitoring and navigation, often outside traditional clinical settings. This shift is not just about consumer frustration. It's reshaping the healthcare stack itself. New companies are emerging to help people find care, price services, interpret results and manage health over time. AI is beginning to play a role in triage, navigating diagnostics and clinical workflows, even as regulation and reimbursement continue to lag behind consumer demand. At the same time, core tensions remain unresolved. Emergencies don't disappear when care moves outside. Insurance costs shift across the system. Questions around access, equity and sustainability become harder to ignore as healthcare fragments into parallel pathways. Today, we examine how consumer behavior, technology and economics are converging to reconfigure US Healthcare. We discuss why insurance defection is happening, how cash, cash pay and proactive care are expanding, what founders are building in response, and what this transition could mean for the long term structure of care. A16Z health and bio partner Jay Rugani sits Down with Nikhil Krishnan, founder of out of Pocket, to explore how healthcare is being rebuilt piece by piece outside the traditional system.

1:02

Nikhil, what's happening?

2:55

What's up, man?

2:56

So healthcare is off to a hot start in 2026.

2:57

JP Morgan always kicks us off in a good place.

3:00

The group chats are buzzing, of course. OpenAI dropped some new healthcare news. They got a health app.

3:03

Totally anthropic. Too.

3:09

Anthropic. Two PHRs are back. President Trump announced the great healthcare plan today.

3:11

I haven't read it yet, so please don't ask me any questions.

3:17

I was going to quiz. I was going to quiz you on that. Utah is letting AI prescribe medications. Of course that's happening. FDA rolling back regulations on wearables.

3:19

Trials. Wearable. We're doing Bayesian trials now.

3:30

We're doing it all.

3:34

Crazy.

3:34

And we have a new food pyramid.

3:35

We do.

3:37

So a lot to talk about.

3:38

Like a downward arrow now.

3:39

Yeah, yeah, yeah. Pyramid. I ran your 2025 predictions through Claude, of course. You got a 7.623 out of 10.

3:40

Already for the year. It's only been two weeks.

3:50

Okay.

3:53

I put last year's predictions. It's pretty good.

3:53

It's really good actually. It's too specific to mean anything totally. But it's good. Nice. So it sounded good, but I thought you went spicier this year. I thought your 2026 were spicier. So we're here to debate them.

3:56

Okay, let's fight.

4:07

Let's fight right now and then we can check the Kalshi prediction market, see who won.

4:08

We're gonna put money where this is. It's not a predictions post anymore. Unless you have some Kalshi, you know, some money. In a prediction market, if you don't.

4:14

Have dollars at stake.

4:20

What are you even doing?

4:21

Why are you even doing it? Yeah, exactly. All right, so let's start with your prediction on health insurance defection. I've got them all up here.

4:22

Cool. Yeah.

4:29

I think that sets the stage on where we think healthcare dollars will flow in 26 and then will flow into the other. So one of your predictions says that uninsured rates explode.

4:30

Sure.

4:40

The uninsured rate explodes and the second order effects. So people refuse to play the health care insurance game. Defecting from the system.

4:41

Sure.

4:49

The rate of uninsured people will skyrocket to 15%, leading to consequences like flooded emergency rooms, spiking insurance premiums, and more cash pay options.

4:49

Exactly.

4:59

So today the uninsured rate is around nine and a half percent. I had to look that up as low as 3% in Massachusetts, 18% in Texas.

5:00

I learned something.

5:10

Yeah. In 20, in 2009, 2010, pre Affordable Care act, it was around 15%, which is 15, 16%, which is where you projected it. So 15 years of progress unwinded in 12 months if you define that as progress. And the whole country is about to look like Texas. So lay that one out for us. What made you think that?

5:10

Well, I mean, it's a combination of things, right? Like at the time when I wrote this, obviously premium subsidies were still a big debate. So it was unclear where that would shape up. And where I live in New York, for example, we have a extremely high cost insurance individual exchange market. And this for a whole host of reasons that we don't have to get into. But it seemed like the premiums were going to spike across the board for the individual exchange. On top of that, small group insurance is kind of going through this weird thing right now where one, if you have relatively less sick people, you're like, hey, why am I subsidizing the pool of the more sick people? Let me like exit it through, you know, things like level funding or any of these other things, or even just not offering insurance at all, right? That's now an option because the labor market has so far tilted to employers that, you know, if you're below 50 people, you don't have to offer insurance either, right? So I think a lot of people with healthier, A lot of employers with healthier employees will be like, hey, I don't want to subsidize these sicker people. And then the employers with more sick employees then are left in their own pool. Those premiums skyrocket and then the employers in that pool also have the option to not. So basically I think there's a lot of reasons why people will eventually just say, hey, should I really be paying? You know, I don't remember what the average premium is, but you know, 12,000 plus dollars a year for a family with a deductible on top. I don't even use health insurance at all, right? People are getting squeezed obviously, and you know, their individual wallet share and all that kind of stuff. So they're going to be asking themselves. And if you go on the, like our health insurance subreddit, extremely common question now of just, hey, you know, should I just not get health insurance at all? Right? These are my premiums, this is kind of what I'm paying, this is kind of the drug I need or blah, blah, blah. And for a lot of people that answer is actually probably no if you think about it for a second. If I were to be really rational about this and I'm thinking, okay, okay. I'm a relatively healthy person with no expected procedures or medical things this year, like nothing planned. If I believe that, then. And my premiums, let's say on the individual market are 550amonth, which is in New York, plus on top of that probably like a 5k 6k deductible right. Now suddenly like you just do the math. Right. It's 12 times 600 plus a 5k deductible.

5:30

Yeah.

7:55

So before even anything is covered.

7:55

Yeah.

7:57

At that point. Right. You are paying like 10, 15 grand depending on.

7:57

Yeah.

8:01

Who you are. And I'm really just protecting against catastrophes, cancer, car crash, whatever. Then it's like expected probability of that thing happening is like relatively low. And so I can totally see for a lot of people saying it's just not worth it for me. So.

8:01

So let's talk about the likelihood that it's going to happen. But almost just rewind the clock a little bit.

8:17

Sure.

8:22

Is this a good thing? Lay out the arguments for both sides.

8:22

I mean the con is sort of obvious, which is that I think a lot of people who are not good at guessing their own risk will have something happen to them and then end up in a hospital and have to pay a ton of money that they just don't have. Right. So that's one con, is that coverage for people goes down and something bad happens to them and then they have to deal with it. Right. So that part's bad. And also I think that depending on what your view is on, should we have health insurance at all then? If you believe that we should have health insurance in health insurance industry in general, then this totally fractures like the risk pool, makes our economics all wonky, blah, blah, blah, and will probably just lead to a handful of payers that do the whole country four or five. Right. You're already seeing this right now. Smaller health insurance companies getting bought by larger ones like happening at rapid scale. A lot of the blues plans are consolidating all this kind of stuff. So if you believe we should have a competitive health insurance ecosystem, this kind of defeats that purpose a little bit. The pros of it are if you believe more in a consumer directed health care system in some capacity where hey, people get money and they should just use that money to pay for services they think are worthwhile to them, then this is sort of the like means to do that. Right. If people feel More health care costs directly. They will be better shoppers. They will, people will try to cater to them more. So there's a better experience and all that kind of stuff. So, I mean, I think a lot of people probably, you know, rightfully make the argument that like having a third party payer is one of the reasons that costs are so inflated. It's one of the reasons the experience is so bad when you're not the direct customer. Right.

8:25

Yeah.

10:09

So those are the pros and cons of it, really. Just, there's no right and wrong. It just depends on what your ideology and what coverage should look like. Is your earlier point about Affordable Care act before and after? Right. Like one of the beliefs was we should have a competitive health insurance market on the marketplace. And they will. If everyone is on the same marketplace, then they'll compete for people and it'll function like a normal marketplace. For a whole host of reasons that didn't happen. So now the question is, should we go back or should we try and double down and fix it? And. Yeah, not clear.

10:09

Yeah. I think there's. If I were to steel man, the libertarian case, which is sort of what you talked about around, around the free market. Let the consumers kind of choose. It's basically saying some people will say, finally, the market is working. People are defecting because the product is bad. Let them defect. The product will get better. If they want those customers, they'll get them back or they won't. And people are voting with their wallets.

10:39

Sure.

11:04

So that's one side. The other side is basically the, you know, the free rider counter argument. We kind of heard this back in 2009, 2010, ahead of the Affordable Care act was people would go and say, I'm going to go without in health insurance, knowing that there's a federal law that effectively requires hospitals to take care of them.

11:04

Yeah.

11:24

And you know, the belief or the, or the worry was that people would go out, go, would sort of stay away from the traditional care, preventative care, and just wait until things get really bad and then they end up in the hospital. Paid for.

11:25

Yeah.

11:38

ER is the most expensive side of service. It's the least affordable way we can deliver all kinds of health care. Yeah. This increases cost for everyone. Hospitals have to pay for it and the uninsured effectively result in all this uncompensated care that we inadvertently pay for anyways totally in higher prices. And so overall, you know, it makes the system work.

11:38

Sure.

11:58

You think we're going to have the free rider problem again with this or is this time, does this time feel different?

11:59

I mean, we'll probably still have that problem, right. If people don't have insurance and they get sick, it's not like they're gonna stay home and just ride it out. They're gonna go to the emergency department, obviously. Right. So we're definitely gonna have that. I do think that, you know, there's like a bigger. I think the main ideological question is like, how do you think we as a country should treat catastrophes in healthcare? Right. I think a lot of the stuff that people talk about when they're like, hey, people vote for their feet or wanna go is shoppable, plannable things. Right. And that totally makes sense. But what do you do with a person who gets in a car crash? Right. People always make the argument of health care can't be shoppable because people aren't going to shop for an ed, you know, they're not going to be on the emergency department shopping for things. But that's a really specific case.

12:05

The highest acuity.

12:52

Exactly.

12:53

Really bad, expensive stuff.

12:54

Yeah. And so, you know, if you look at other countries, models, for example, right. Like Singapore, maybe like the most common example here.

12:55

Yeah.

13:01

They try to separate those. Right. Like, they try to make shoppable services more of a cash pay experience with their, you know, their version of what looks like an hsa. And then the government covers more catastrophic things, Right. Cancers and inpatient and all this kind of stuff. Right. So maybe actually this is us speed running to a system like that where you're like, okay, hey, we got a bunch of cash pay patients. They're also about to show up to these higher acuity places and we need to figure out a way to cover them and we need a scheme to do that. Right. In some capacity.

13:02

So you could also argue that the goalpost has moved in the culture around health. There's clearly a greater cultural movement around health. Right now we've got maha. Consumers are spending more out of pocket for health. Alcohol sales are down, supplement sales are up, you know, and so we're not.

13:33

Drinking alcohol, we are injecting peptides.

13:54

Yeah, exactly.

13:56

We'll get to peptides.

13:56

We'll get to peptides. You know, one question that, that I think a lot about in comparing this to the kind of free rider problem that people talked about a lot in 09 and 2010 is are these actually different populations? There's one population which is the proactive health crowd, you know, higher income, they're buying supplements.

13:57

Sure.

14:19

They're on, you know, they're buying GLP1s out of pocket. They've got an aura ring, a function health membership, a council membership. They're doing all the things, all your friends, they're working with everyone. You know, they were already insured and they're likely going to stay insured, but then they're doing these extra things out of pocket. And then there's the newly uninsured that you're arguing is sort of accelerating. People are losing coverage because subsidies expired because the big beautiful bill had Medicaid cuts and now they can't afford it anymore. And we're not doing a good job as a country taking care of those people. What do you think of that distinction?

14:20

I mean, I think at the end of the day, everyone wants to be healthy. Right. Just the way they define that is a little different. Right. I think there's a group of people probably again on the higher income scale who are like, I don't feel like I'm being taken care of because there's, you know, I'm not being monitored, for example. Right. Like, the health care system is sort of not designed for my view of how I think healthcare should be. And so I am opting out into a separate system that I think more aligns with my view, that is more consumption, you know, like just, I want to be able to consume more health care and this is like the way to do it. More lab tests, more MRIs, whatever. Right. I think if you are lower income, you still obviously want to be healthy. No one like that, that is still the goal. But I think the problem set is a little bit different. Right. Which is I don't want to have to take off work to do a doctor's visit for just a prescription refill, for example, or my kid is sick and I just need the answer to what is happening here. Right. Or the drug I normally take costs X amount. Now how do I get it for cheaper? Right. I don't think that's maybe the problem set is a little different, but the end goal is still the same. It's just like the way they think, the way people think about health care, about what is good health is just going to be, you know, different between those groups. But they still want. Everyone wants the same thing. They want health care for cheaper or actually, you know, one thing is that I think is very common is everyone wants more transparency on the pricing. Yeah, right. There's like, hey, I think I need to go to the hospital, but I can't do $100,000 bill. Right. Or I want to get lab tests Done. But like no one is telling me how much it costs if I like go to the hospital to do that. Right, right. So everyone wants price transparency and I think the cash pay market on both ends try to serve that.

14:57

You think there's going to be a third category that emerges here between those two camps? As an example, a member of the middle class who could in for, who could afford insurance but chooses to go uninsured, they probably would have got health. They probably either work for a very small employer that maybe doesn't offer health care or they, you know, work for themselves and they would have bought on the exchange as an example and they're going to bet on cash pay plus proactive health and risk it because that would be a third category in between these two. Do you think that's where the growth in uninsured comes from?

16:39

So you know what health sharing ministry is. So this is like the explains, okay. So for people who don't know what this is, it's basically instead of health insurance as you know it, which is, hey, I pay premiums and into this pool. And then when things happen, they pay out to other people. In health sharing ministries, the, you know, high level is you preload a wallet essentially with something that looks like premiums, but it's technically not premiums. And then when other members have a hospital bill that happens, people contribute from that wallet to pay for the hospital bills. So it's a little bit more of a crowd, almost like crowdfunding for healthcare, but amongst a specific pool of people that are essentially pre vetted before they come. So those are typically much cheaper. And there's for a bunch of reasons. Right. One, they can do things that plans used to be able to do pre Affordable Care act deny people based on pre existing conditions. Right. So their risk pool is much better. Right. As a whole. And also they don't have the same regulatory requirements of hey we, you have to have a bunch of capital on your book so you don't go insolvent. This happens. Right. So your regulatory, your regulatory requirements are lower. So your cost of for compliance and all that is lower too. But it's like an interesting model where you're like, hey, if I go to the hospital and get a bill and I show this to the people and then they all contribute to it. But the downside is you got a better, you better hope that people are actually down and will pay for it. And I do think it's an interesting product for people who are. And when I say interesting, I don't mean like I ideologically agree with it.

17:15

But I do just think you're not recommending.

18:44

I'm not recommending it, but I do think this is. It probably addresses this middle ground a little bit where it's like, hey, like, the health insurance I currently have sucks and it's not covering the things I wanted to cover anyway. And it's really expensive. I'm already gambling on whether or not the insurance is going to cover me or not, which is the case for a lot of people. They're like, you know, they. I'm buying this. So then I go and they deny my claim there already there's a chance of that happening. I might as well try this other thing. And if I can get in because I'm a healthier person, I pay less for it. So I think there's a lot of people that also just ideologically agree with that premise. So that I think becomes sort of potential for people in that middle zone. Yeah, but the infrastructure is not built for that because a lot of those people, again, have to go and shop for their own care, negotiate the bill, all this kind of stuff, which. And some of these health sharing ministries offer services to help you negotiate and all this kind of stuff. But it's just also a very different model of health care consumption. Like, it's. There's also some cultural change that happens here where people have to go and are willing to like, negotiate prices with providers. That is not a normalized thing. And maybe that becomes more normal.

18:46

That makes sense. It's interesting.

19:55

Would you sign up for a health sharing ministry?

19:57

I don't know. I don't know if I would, actually. So I don't want our health care system to splinter into two groups. One group that can afford to defect from health insurance, go cash, pay and run their own healthcare stack that way. And then folks that are effectively cycling through our existing healthcare system, you know, relying on charity care effectively out of pocket for the health system to pay for it. I spent a little bit of time this year studying different healthcare markets. Mexico, as an example, has that to some extent. It's not a perfect parallel where there's one system for the rich and another system for the poor.

19:59

Yeah.

20:35

So that would be the. That would be the downside case that I would worry about here.

20:35

There is also, I mean, so for one, I think the core problem with US Health care is just like, we have too many risk pools. Right. You kind of have to choose one or the other. Like having this, hey, maybe we're a little cash pay. Maybe we're A little bit pay, a little bit insurance. Some people are in employer insurance or some people are Medicare. That's kind of the core problem. Right. We have to move to one or the other. Right. So, you know, for the aca, the idea was we'll move everyone theoretically to this individual exchange and everyone will buy from that. That's at least they're all in the risk pool. Right. And that looks more like Germany.

20:39

Yeah.

21:08

You could also do totally cash pay, right. Where it's like everyone is cash pay and you're just focused on that. That looks maybe more like an India, for example, or like a Singapore, if you're doing it through splitting outpatient and inpatient out.

21:08

Right.

21:20

But you have to pick one or the other like this, like in between.

21:21

Makes sense. Americans love choice. We love choice. We, everyone wants to have an option, we pay for the choice.

21:24

This is. Choice is like the worst way. It's like, it's like it's choosing between the worst options now. Yeah. But also on the flip side, there's again, like the ideological question of do you believe that you should be able to pay for better care?

21:30

Yeah.

21:45

Right. And I think a lot, that question makes a lot of people really uncomfortable because it's like they, they cannot reconcile internally that, oh, people can, should be able to pay for better doctors. Well, if you don't want that, and that's fine. But what ends up happening is that if everyone gets access to one care system, then there will emerge a parallel system that allows them to do that no matter what. Right. Because they'll say, hey, I'll pay for the best doctors, cash, even if there is an existing one. And those doctors will exit. Right. And that's kind of what's happening today with a concierge medicine and all this kind of stuff, right? Everyone's, hey, we want primary care physicians to be able to see everyone. It's like the great equalizer, all this kind of stuff. And then suddenly people are like, I will pay a lot of money for you to just not do this and then come just take care of me. Right. So the reality is that's kind of already happening. I can't think of a single system. Even in socialized medicine countries like the uk, people still get private insurance and to see, see docs with lower wait times, all this kind of stuff. Like, I don't think it's possible to actually have a system where you can't pay for better care. It just happens, it'll just emerge as an economy no matter what. Unless you, unless you ban It. Right. Which like you can do. But that has its own sort of downsides to it too. Like Canada, for example, is an example of not being able to a parallel system. And that has its own issues where people are like, hey, if I have a complex care thing, I'm going to come to the US So I don't know, like we got to get comfortable with the idea that I think you have to, you will probably, no matter what, end up with a system where people pay for better care. You can either do that in a really convoluted way or just do that in a straightforward way, or do that in a way where you're more comfortable with what they choose to pay for. Right. So for example, in countries like India and stuff, you can pay for better amenities. Right. Like the nicer wards and all this kind of stuff. Like maybe that's a more palatable way to do that basically.

21:46

Yeah. Yeah. It's interesting. Where is the. So if this plays out and you know, TBD if it's 15% or 18% or 12%, but it's clear that, you know, per the Reddit threads, which is truth.

23:40

That's my truth.

23:54

That's my truth. People are defecting.

23:55

Yeah.

23:57

What are the ripple effects for startups if you're trying to start something? You know, obviously the proactive health startups have an opportunity here. More cash, pay spend. But what else?

23:58

Yeah, I think also just like care navigation looks very different for this group of people.

24:07

Totally.

24:11

Right. I think you have to really think about is this thing I'm doing serious enough that I need to go see a higher cost medical practitioner if you're paying out of pocket. Right. That triaging step is really different. So I do think that helping people figure out like, hey, where do I go for this? And is this actually serious enough that I need to go becomes a whole separate type of product to build. Basically compared to someone, if you have insurance, you're really navigating them through their insurance directory coverage, whatever it is. So I think that looks very different. I think another ripple effect too is there's going to be, I think a lot of people that try to deliver care at like hyper lower costs. Right. So the Doctronic, you know, pilot in Utah is a good example where I, I remember correctly, they are doing prescription refills at $4 a pop instead of 150. Yes.

24:12

Yeah.

25:06

Which, you know, that is actually pretty. That is relatively affordable. You need a prescription refill.

25:06

Yeah.

25:11

That is like one route you can go rather than having to do a net New visit to the doctor. Right. So I do think you'll see more of this. Hey, how do we bring costs? Like an extremely low level, obviously using AI, using tech, using all this kind of stuff. And then the other one, which like I think is a little interesting to think about is, you know, can you do some form of novel contracting with the existing providers for when they see people who are coming in as cash pay?

25:12

Yeah.

25:40

Right. Because the reality is the providers are going to be in a tough spot. Right. Where someone comes in, they have to pay cash. They don't like, you know, if they come into the emergency room, they do some really expensive stuff, people leave, they chase them down for the money and then they don't get it. Right. So the provider is in a tough spot when that happens. Can you offer really easy like bundled payments or whatever it is to people to be like, hey, if you're a cash pay person, here's what a rate is this and come through and get it right. Now if you wanted to figure out the cash pay rate for something, you like have to go through a really weird convoluted process to do this. You call in, you'll say you're self pay patient. You're kind of like haggling. I feel like I'm like the, in the, in an Indian market, you know what I mean? I'm just like, you know, we're haggling back and forth and we're just like, what's up? Where are we going to be? But to maybe just make it really easy for everyone's like, hey, here's the bundle here are the services included. You can come for this. And that's going to especially be true for things like somewhat higher cost procedures and stuff like that where it's like if you're uninsured and you do really need that thing, here's how we can do it and here's the cost we'll pay.

25:40

It looks more like the medical tourism market where people are coming.

26:46

Exactly. More.

26:50

But it's a local version. Yeah, but people are coming to the US and they want to get the procedure at Mayo Clinic.

26:51

Yes.

26:57

They're paying out of pocket. They're not in our system. Yeah, there could be a domestic option like that. It exists today, but it just make it easier.

26:57

Lower friction. Very. It's very hard to do that.

27:04

Yeah, yeah.

27:07

So maybe that's one.

27:08

Okay, so we're going to jump around across some of the predictions just to hit ones that are more relevant. If consumers are defecting, coordinating more of their care and spending more out of pocket. One prediction is that screening diagnostics and at home sample collection becomes the hot area.

27:09

Yeah.

27:27

So, so lay out the case there. So little John is doing colon cancer screening commercials. People are shipping their poop around the country like just crazy for health care. Not even just doing it, you know, check your, wash your hands after you check the mail. Lay out that. Why do you think that's going to happen?

27:28

Well, I mean to be clear, I don't think it's because the uninsured thing happened.

27:44

Sure.

27:48

But I think that dying diagnostics is, I feel like such an interesting area that is like very underinvested in. You know, I think maybe one thing to tie to the uninsured, maybe not uninsured, but cash pay rather is that I think this is one of the areas that a lot of people, when you talk to doctors and you talk to patients, this is one thing that I feel like there is a somewhat irreconcilable difference here that they just can't think about. Like the people want to be monitored in some capacity and a lot of the doctors will be like, well you can be monitored but there's not much we can do even if you get monitored. Right. Like it's not like your care pathway is going to change during that. But a lot of people just like want to be monitored. Right. Because they feel like someone is looking out for them and they're being considered at all points and blah blah, blah. So I think that people want that and especially in the cash bay and you see this with the longevity medicine and all this kind of stuff, like a big component of what they want is just like being monitored. Right. Even if it says nothing, it's just at least we checked, you know. So I do think that the appetite for that is really high. I think as you move to more of these AI native care models, that becomes a really core part. Right. The shift from reactive care to proactive care really has to come from what is the data source that's going to tell you that something has to be done. Right. And I think that is going to come from some form of screening, diagnostics, et cetera. You know, we talked a little bit about the AI, about AI prescribing for example. I think the next level above that which will be interesting is like AI ordering things like labs or follow ups that you might need pre the doctor actually needing to look at it. Right. And I think that is going to be sort of a part of this. But in general I think diagnostics also are Seeing a more interesting go to market strategy too, which is hey, we basically see a big cash pay market that was just wanting to be monitored while in the meantime maybe we go through a more legit, you know, lab developed test or FDA approval route. But there's this appetite here for these like new tests should just be monitored on different metrics that maybe are not typically monitored and they see this appetite from people to do that. Like especially for example if you have, you know, some family risk of something. Right. But you don't quite fit really neatly into guidelines around like that you need to be prescribed this thing. People are going to find alternative routes to doing that and they have money to do that too. They're going to pay for it. Coronary, you know, calcium CT scans, like great example of this where like for some reason it's really hard to get that. And also there's still a huge lag between like evidence and sorry, research and then care guidelines. Right. So the lag is like 10 plus years in a lot of cases between what's interesting and what actually makes it into hey, we're doing this for all patients and all this kind of stuff. But people are willing to pay to escape that.

27:48

Have you done it?

30:40

Calcium? I'm not dude, but I'm like, you know, if you're a 30 year old brown person, like you're suddenly in a risk category that you have to be thinking about this.

30:41

I'm a perfect example of what you just described, which is slightly high on the border cholesterol, South Asian male. So I have risk factors.

30:53

Canon event.

31:02

Yeah, exactly. And so my doctor was like, you definitely don't need a calcium CT scan, but what do you know? Doctors know. Just kidding. But I did it anyways.

31:02

So what'd you find out?

31:11

It was a zero, which is good, but I wanted to know that.

31:13

Yeah, it's good to know exactly. Like I, I think the tough thing, I think the thing that we have to figure out as a society a little bit is that the existing health care system really wants to be like, hey, here. They really want to create standardized care for people, right. And standardized guidelines to make medicine also good at the median level for everybody. Right? But at the same time people want more agency in their health care in some capacity, right. They want to feel no one wants to be told, like, hey, just wait and see. They just don't want to be told that. And there are probably a lot more lower risk things that we can just let people try and do and see. Like I think this is one of the reason, for example, that protocols are so popular right now, like Huberman protocols and protocols, you know, the core, the core thing of what these protocols are saying is eat well, sleep well, you know, normal advice you get.

31:15

But it's branded.

32:10

Well, it's branded, but also it gives people like what seems like a task list that they should actively be doing to like be good at their health and gives them like supplements to feel like they're doing something.

32:11

Yeah.

32:24

And whether or not they're doing it is sort of irrelevant. But the feeling of having agency in those situations I think is actually the important part. So the same with the testing. People just want to feel like they're in control a little bit. And you know, I think there's some tension here about, okay, there's clearly a line of where patients are just like YOLOing until screening tests and diagnostics that then end up with like more pressure on the health care system. Like you find incident dilemmas in your mri, now someone else got to deal with it, all this kind of stuff. But I think people want more agency and getting a test done where you're like, I just don't feel good about what my care plan is today. I just like to see and test is just for better or worse, just going to be a part of healthcare going forward.

32:24

I think that's right. And I think that creates another debate which is what do regulators allow these tests and AI applications to say? Yeah, so I think a lot of these things, you know, the medical establishment will say, well, it's not really going to tell you anything that actionable or that interesting. Yeah, but these things start on the fringe. My take is that I think it's a good thing. They start on the fringe and then they get more mainstream. I think the reality is that the healthcare system that we have today in the US is drastically supply constrained.

33:05

Sure.

33:34

Yeah. So 23 million Americans work in healthcare, yet we've got long wait times. You've heard all the statistics. 100 million people plus don't have access to a primary care doc. 40 day plus wait times to see a doc if you can, etc. Etc. And so all that friction is frustrating. And the demand exhaust shoots off somewhere and it's shooting off to all these places. And to your point, people want agencies so that agency is going to go somewhere. And where is it shooting off? It shoots off at LLMs. So you saw the OpenAI report. 230 million people around the world using ChatGPT for health related questions each week. 40 million a day. It's crazy. It's crazy. It's replaced, you know, Dr. Google and many other things.

33:35

Yeah.

34:26

You know, the proactive health tools are being adopted by a bunch of people. The health memberships, the wearables like Aura and Whoop, you know, name your thing and you know, that makes sense.

34:27

Right.

34:39

People want to do stuff with that. But there's still regulatory friction that I find interesting and I'm kind of curious to see how it plays out this year, which is at least as of a few weeks ago and prior to that there were two buckets for a device. FDA either said it's a wellness product and it can't make any medical claims at all, or it's a medical device and you can make all the claims you want, but you've got to go through a long regulatory approval process that, as we know, largely favors the incumbents, the big med tech companies really hard for startups. So there's not that many on a relative basis, devices that have emerged relative to, I think what could have been possible.

34:39

Yeah.

35:17

You know, I'm touching my OURA ring as an example. So, you know, so Tom at Aura, we're not investors there, but I think he wrote this amazing op ed in the Wall Street Journal that basically said there should be a third category. And that third category is digital health screeners, which is there's something in between that's not necessarily going to give you direct medical diagnosis, but will show you a couple things that you probably want to know about your body and you might want to talk to your doctor about. So, you know, the examples that he gave in the article that I thought were good were maybe there's a couple early indicators of high blood pressure that it can infer and tell you to say, hey, you might want to talk to your doc about hypertension. And I think I can't remember the last statistic, but 40% plus of adults are diagnosed with hypertension or something crazy like that. So it's a real chronic condition that's challenging. The other one that they give in the article is, you know, hey, there's some sleep irregularities that are probably predictive of sleep apnea. Talk to your doctor.

35:18

Yeah, totally. Right.

36:14

Like we should just let that go. And I think that's the. There's obviously something in the middle where you could get all this data and then just drop it into your LLM. People are doing it anyways. I drop all my medical data into Council and you know, pick one.

36:15

Yeah.

36:31

And you know, that's where it's going. FDA said some things that suggested that they're going to.

36:31

They give the hint. Hint, wink, wink.

36:39

Yeah, they're going to, they're going to roll back some of these things on wearables. But like wearable is just the start. So right now you get into AI doctor applications and what they say, what can they do? So what do you think of that?

36:40

I mean, again, I think the question is how much as a society, how much risk do you think that patients should be allowed to take on themselves? Yeah, right. Like the AI prescribing thing maybe is a good example here. Right. And we'll come back to the diagnostics for a sec. But do you think that a person should be able to say, hey, I want to be able to pay $4 to see the AI doctor who's going to auto prescribe me via these common meds? It might get it wrong sometimes and if it does, then it'll harm me. Or I can choose to pay a higher amount and see a doctor that maybe has more sort of higher accuracy or whatever it is and lower chance of me doing risk of having an adverse event or whatever, like getting prescribed the wrong thing.

36:50

Whoever.

37:34

Should I be able to make, Should I be allowed as a person to make that choice that I want? The AI doc that maybe gets it wrong more often, but is cheaper or not? Probably same thing for screening diagnostic wearables, right? Where it's like, hey, here's a thing that might find some interesting stuff in your data, but also it's going to have a false positive rate of this much. Right? And you can choose that. Now I will say the issue here a little bit is that the burden of the false positive rate then falls on the doctor you go and see or the health system or the emergency room. So I think that is like a real consideration we have to think about is like when people have these devices at scale, suddenly the risk of when false positive rates happen, it's gonna be millions of people. Right? That's no joke. Right. But also at the same time, I think we're missing a lot of really interesting data that we should be doing more things with. For example, I think been talking to a bunch of people that have been wearing CGMs for a while and a lot of them, I think, for example, have different pathologies of what traditionally looks like diabetes. Like, for example, I think diabetes is this like broad category, but there's probably multiple underlying things that are actually happening that are actually different forms of diabetes. For example, probably same thing with like sleep, insomnia, all this kind of stuff. There's probably underlying multiple pathologies, but it's hard to tell that unless you have a more massive stream of data that is coming that can help you tell like how that's changing and how that's changing from a healthy baseline. Which I think is really important. Right. I am wearing the OURA ring today as a healthy person. As I develop a disease, it's going to catch things that change from my healthy baseline to like being a sicker person and those changes in how that naturally progresses, like natural history study basically is really important and I think we probably can do that for more diseases if we had more people wearing this kind of stuff.

37:34

So this is the real world data dream. It is really what we all wanted.

39:30

Yeah. And I think, and I think the problem is real world data has always been focused a little bit more on like pharma use cases. I think there are a lot of people that are also just like, I kind of want to know if there's other people like me who have this sort of weird sleep pattern.

39:34

Yes.

39:47

And what is that? What are they doing? What does that mean? You know, that's I think a really open, unexplored space that people have not quite, you know. Yeah. I think if anyone is listening to this, I think building a patient first biobank on these concepts I think is really interesting. People really do want to know how people that look like them or have biomarkers that look like them deal with that or operate in the world. What does their care journey look like? Even if you de identify it or all this kind of stuff, it's useful to know, hey, this person who had this weird sleep pattern took this magnesium dosage and it actually seemed to work for them. And they actually, you know, when we sub cohort the populations, they look very similar to you. So maybe you should try it. Right. That's very unexplored. And I think wearables and all this kind of stuff is a way to do that. But we can only do that if there is some way to capture that data in a structured way, basically. And you need a middle path to do that.

39:47

So that's a good transition to talk about. You had three or so predictions around healthcare AI and your first one was around state governments clashing with the federal government on key regulations. Talk about it across a few dimensions, but let's zoom in on AI. So state governments will clash with the federal government on public health guidelines, vaccine recommendations, AI regulation, insurance law nuances, all the above. Yeah, let's talk about that. So on AI specifically I think this is interesting. Last year we had Illinois and Nevada and a couple other states either ban or restrict.

40:42

Yeah.

41:16

AI therapy, chatbot usage basically last week. So since you wrote your piece, we literally had Utah saying, doing the pilot around AI being allowed to prescribe medications. It's very narrow on 190 low risk medications, your asthma inhaler, you know, things like that. Low risk stuff to start, but can expand. And then President Trump signed an executive order in December saying that effectively said, hey, we're going to do a national framework and we'll make it easy for folks to follow. It sounds like you think it's going to be more chaotic in 2026. Definitely say more.

41:17

I mean, I think it's, I mean, I think there's a couple things, right? One part is there is definitely an anti AI sort of movement that is happening in the US which is sort of expected and natural. And people are worried about job loss and they're worried about how decisions are being made that affect their lives because maybe there's an AI on the other end that is choosing things that are anti, helping them. So I understand where that's coming from. And that anti AI movement will inevitably impact policy. And depending on which state you're in, you are going to have a regulator in that state. Side with one side or the other side, maybe like a analog here is actually like Waymos. Right? Waymos are in a few states. They started in California. They are super like I love riding in a Waymo every time I come here.

41:56

Yeah.

42:50

States have very different opinions on whether they should allow Waymos into their own state. Right. Whether it's, they care about how much they have to prove safety wise before they roll out or they, you know, there's, they know there's gonna be widespread job loss when they come in. And I think AI for other things is going to follow the same trend. Right. And so I think the issue is that like states don't quite know what to do. I think we can all acknowledge that there's a lot of also bad AI use cases that we need to watch out for, using deep fakes in weird ways and a lot of this kind of stuff. Right. So as a whole, I think people are trying to figure out how do we regulate the downside risk of AI while also enabling the upside potential that we can see. Right. And maybe it's actually not a bad thing that we're running a bunch of state by state experiments to see what works and what doesn't and then create a more national framework to do this. The downside though is then you have to. Healthcare has always been really annoying about having different compliance rules for every state, which means your product looks very different and the cost to go in a new estate is very different. That's gonna be annoying. But actually, maybe we should be running different experiments here. And each state is gonna, you know, it's gonna look different depending on their own healthcare needs as a state. Right. If you have a really large rural population, for example, that really struggles to get access to care, you are probably gonna lean more heavily into, hey, we need to get. These people don't have a pcp. We need to get them something. Right. If you have a very competitive provider network, maybe that's actually not the case and care accessibility is fine and all that. So I think that'll. I think it just really, I think that states are trying to figure out, like, how do we protect the downsides while getting the upsides, but they just don't. They all have a different opinion what that means. Like, some states, for example, you know, they really don't want insurance companies to use AI to deny claims. Right. They're like, they think that's like really bad, blah, blah, blah.

42:50

They'd rather have humans deny.

44:46

Yeah, yeah. Which, which also, at the same time, I'm also like, how are they even going to monitor, like, if a human is using ChatGPT to lead the claim and then deny it? Is it really any different?

44:48

Yeah, sure.

44:56

Not to be clear, but I don't know, we'll see with some fun home state experiments.

44:57

Yeah, I think that's right. I think from a competitiveness standpoint and from getting all the innovative ideas out there, which ultimately benefits the consumer, I think you've got to have a national framework.

45:01

Sure.

45:13

Because if you have a lot of regulatory complexity with different rules in every states, it. It favors the incumbents who have the armies of legal teams that can interpret all those things and operationalize and scale that way. You want to, you know, level the playing field for big tech and little tech. On the other hand, you want the experimentation. I think it's. I think when the technology is moving so fast, you want regulation to catch up, but you really don't know what the right answer is. And so it's instructive to have different states experiment for their unique population mix and rural to urban mix, as you described. And so I think the question for me is how do we have the right data infrastructure to learn from all that stuff? So, like the.

45:13

Sure, that's true.

45:59

The prescription example is an interesting One. Okay, great. You know, you can refill a prescription for $4 instead of $150. Let's see if that works and let's see the upsides and downsides of it. We're going to learn a lot from that state pilot.

45:59

Sure.

46:12

But that's just one example. We want 50 different examples to play out. And that's the thing that I worry about.

46:12

Yeah.

46:20

I mean, you know, if we wait just for national. Basically.

46:20

Yeah. I mean, there's an argument to be made also that I thought actually that I think Jonathan Slotkin wrote like a great op ed about Waymos. Seem to have proven that like they're so good, that actually would be unethical to not rule it out everywhere.

46:23

So good.

46:37

Right. And I think if you can. If we can do something around in the lines of, hey, this pilot or this thing looks so good, we really just gotta boost it everywhere. I think that can actually be maybe a different way the federal government thinks about things. Rather than be like, hey, we're going to set blanket rules. It can be like all these places should test their own things out. If something looks like it's working.

46:37

Yes.

47:00

Then we make that the rule for everyone else. And then like Speedrun getting people, getting those, getting that, whatever. Getting those features or programs out to everybody.

47:00

Yes.

47:09

Rather than be like, hey, we need to create a yes, it's working. No, it's working. Rule set for every state.

47:10

Yeah.

47:15

Just be like, go run your own pilots and tests and then let's get it up.

47:15

The New York Times article around Waymo is a perfect example. And actually, ironically, it. It pointed to healthcare and life sciences as a industry that already has this framework, which is if you're running a clinical trial and it looks great and the evidence is as clear as day.

47:19

You stop the trial.

47:34

At some point it becomes unethical to continue the trial. You. Because you're giving the placebo to the control arm and it's causing, you know it's harming and you know it's harming them. And so what is the framework like that here? I think this is going to be one to watch.

47:35

Yeah.

47:48

That relates to ip.

47:49

Sure.

47:51

And all the IP debates around healthcare AI.

47:51

Yeah, yeah.

47:55

So you said intellectual property lines will be drawn in healthcare AI with lucrative copyright and licensing deals. This photo, which we'll put it up for the people out there with big birds strangling.

47:56

You know, sometimes my wife walks in on me doing my work and in this case she's seeing me like output an image of Elmo choking me out and she's this like, really, what's putting food on the table?

48:06

She's wondering if you have a real job. Yeah. So. So, I mean, but it's crazy. So. So lay that out, you know, lay out that case for AI overall. You know, all the drama, deals and litigations that happened last year. And then what does it look like in health care specifically?

48:17

Yeah, I mean, on the. In the foundation model side, you know, the big issue has always been like, what does copyright look like in an era where, you know, the AI service regurgitating or spitting out things or like using IP in ways it was never supposed to be, Just like Elmo choking me out. Right. Yeah. And now there's a bunch of lawsuits, slash partnerships that have been happening where different foundation models now are allowed to use the IP from different companies. So I think Disney invested in OpenAI, right. And now presumably can use their IP within it. So, you know, you clearly see these lines being drawn. And Anthropic had a huge case against people who have written a bunch of books and all this kind of stuff. So that's being figured out. Right. Healthcare has so many copyrighted things. Disease scales, for example, CPT codes. Like, so many of these copyrighted things probably is the most behind a paywall. Expensive PDFs that exist of any industry somehow.

48:32

But they're. Are we the market leader in paywalls?

49:27

We must.

49:29

That's really sad.

49:30

We must be. So if you want to use any of those things, you got to pay a fee. You got to pay the piper. Right. As long as we are in a system where you have to use those standards or whatever, it behooves a lot of the healthcare AI applications to create licensing partnerships with these people who own the ip. Right. So that'll probably happen. Realistically, I think maybe a more interesting one. And I sort of posted this online the other day and kind of got roasted. But I think it's an interesting thought experiment is I think protocols are ip, right. I think that the Peter Attia longevity protocol. I would buy ip and I think one question has always been like, okay, if you are a Peter Atia. Right. By the way, I don't know Peter Atia, but hey, man, if you're listening, let me know. Let me know what you think. I would probably not die.

49:30

I would buy the Peter Atia protocol.

50:22

Yeah.

50:24

Because I can't, you know, his. To be on his client list, he charges a really high fee, but make that available. Yeah.

50:24

Like, also, if there's a Peter Attia trained agent that is implementing the protocols at the, you know, patient intake level, at the like dose escalation level, whatever, then you can potentially just see a regular doctor that has a wraparound of the Peter Attia protocols that, you know, helps you do all the things that you normally would at his clinic, essentially. Right. And you pay maybe a lower price than the normal clinic. Right. But I think that becomes an interesting thing of hey, I want to create my own protocols that I think are maybe slightly outside evidence based medicine guidelines and then gets implemented at other clinics.

50:31

Right.

51:09

But a lot of those clinics might implement it in the form of AI agents or other ways to actually execute that. So that's like a slightly different area that I actually think is maybe more interesting. But at the baseline you're already seeing like open evidence partnering with like JAMA or I think it's New England Journal of Medicine just to be able to like pipe that data in. So yeah, all the paywalls will probably end up finding their own specific healthcare application and be like, hey, this is the one we're licensing it to. Whether it's exclusive or not. It's sort of tbd, but at least they could probably have those agreements.

51:09

What do you think are the biggest data sets in healthcare that are still untapped? Because you know, there's builders listening that are just trying to figure out what's next. So ama, CPT codes, Milliman Clinical groupers, health system, clinical EHR data, PBM data, famous doctors with their own protocols.

51:41

Well, you know, the honest answer is I think a lot of those data sets are like not great data sets. Yeah, they're, they were created and built and used for a very different purpose than probably want to use it today. Right. Like today we use so many of those data sets as proxies to figure out something other ground truth. Right. Like for example, we will use prescription claims data sets to figure out if a person has diabetes rather than just seeing like hey, let's connect your CGM and just see do they have diabetes or not. Right. I think a lot of the existing data sets in healthcare are like quite bad actually. And I actually think the more untapped data sets are these net new data sets that are going to be created either from screening wearables, et cetera, but also things like scribes, for example. Right. Scribes capture raw audio text data from the encounter that is like totally net new data set that is really interesting to use that you can do with, you know, lots of things. Like for example, for anyone listening, one thing that I think would be really interesting is net new medical malpractice carriers. Like why is there not now a metro mile equivalent but for medmo, Right. You put a thing in. If you are, you know, we sort of listen to how you practice medicine. If you tend to do it within like appropriate guidelines and all this kind of stuff, we will actually give you a better rate.

51:59

Right.

53:18

I think a lot of these net new data sets will actually allow us to do other interesting downstream things too. But I think the old data sets are just like, they're just kind of jank and used for things that are like not. They're not supposed to be used for. We got to stop using fucking claims data sets for real world evidence. It's just that wasn't meant for that.

53:18

Yeah, I mean especially like phrs are back and then OpenAI bought a phr.

53:37

Yeah.

53:42

And then maybe phrs are not back anymore. Maybe the LLMs are gonna do it. But like personal health people are uploading a lot more. I'm uploading a lot more of my stuff.

53:43

This is a great example, right. Like phrs. Now obviously for personal health records have been one of those ideas that everyone has pitched every year and then. But you know, at some point it'll happen. Like it's just kind of unclear when. And the missing piece has always been interpretability. Right. It's hey, you're giving me all this data, but I don't really know what.

53:51

To do with it.

54:10

I got to go back to the doctor anyway to interpret it. Not that useful right now with AI tools you have more interpretability that unlocks a whole new use case. And more importantly is that it gives patients a reason to now dump more data into these things. Right before it was like, okay, if I dump more data into the personal health record, my doctor doesn't want to see all that. It's a lot of noise. I don't know what to do with that. There's nothing more to do. Right? But if now suddenly I'm like, oh, I'm actually getting more interesting answers the more data I put into it, then suddenly you have patients wanting to generate more of their own data to put into this. Right. That creates totally net new data sets. You know, patients might actually also like you came from the clinical trial world. For example. You know, one part of clinical trials is patient reported outcomes. Those like diaries that patients have to fill. No one's filling that out. Right. It's a terrible completion rate. But suddenly if you're like, hey, you'll actually learn more by chatting with this thing that we have that will tell you more answers maybe about the drug or anything like that. There's now an incentive to actually like chat with this and given more information. And that's like totally net new data, which I think is also probably closer to ground truth of how people are actually feeling about their health and stuff.

54:11

Though that extends to anti AI populism.

55:25

Yeah, so.

55:29

So lay that one out. You know, we basically we effectively just spoke about the bull case for AI bridging the supply gap. We don't have enough docs, not enough physicians. AI is going to bridge the gap. Explain the populist revolt.

55:29

I think it's a few things, right? Like one, I think, look, the reality is healthcare is a jobs program in the U.S. yeah, right.

55:43

23 million.

55:52

It is the number one employer in the U.S. especially. Or I would say these, you know, kind of administrative jobs in particular. Like the number one job category is still patient care. Right? Like it's patient facing caregivers and all this kind of lcsw, all this kind of stuff. But there's a big layer of administrative people who are very comfortable in their jobs, do not want. Can their jobs can definitely be automated and do not want to see that happen. So I do think one thing you'll see is just people afraid of job loss. As with any other new technology as it comes out and wants to fight back against that. The second thing that I think you'll see happen is there's going to be a weird fight that happens between what a doctor thinks they should be doing and when they think that the AI should be telling them what to do. Right. Because this is going to be like a strange thing where either hospital admin or payers or something are going to start saying, hey, you got to consult the AI. Yeah, for this xyz or we're mandating scribes and the scribe is not that good. And then suddenly they're like having to repeat a lot of stuff. Like I do think the implementation of this kind of stuff is going to be kind of clunky. And because of that people are going to have bad experiences with it. And when they have bad experiences, they're going to feel a little bit like they're being encroached on. Right? So I think those are some of the areas that'll happen. And then I also think realistically there's going to be some big AI mistake that happens in some capacity. You're kind of already seeing that with ChatGPT, with the kids who are committing suicide because the bot is convincing them to do stuff. And people are already having some like, hey, we need guardrails here. Right. There's some backlash on what we think should, how it should be allowed to interact with kids and all that. And I think you'll see like a big healthcare issue that happens where someone's really misdiagnosed, really publicly or something like that happens and then people are going to say, hey, we need more guardrails here or we need more observability or we need something that tells us that hey, this is, you know, this is a one off or we won't let it do really high risk things or something. I think people will push to regulate those things. Whether it's right or wrong, I don't know, but I just think that'll happen. And by the way, you see, already saw this with the Internet when it happened, right? Ryan height laws are a great example of this. Right. Where I believe it was a teenager got access to controlled substances through a Internet enabled service of some kind. And then they put rules, and I overdosed, I believe. And they put rules around like what you could do and what you couldn't. And then those rules just stayed for a really long time. It's hard to roll back rules like that once they get implemented. And if the case is high profile enough. So the question is, this will definitely happen. What is our response going to be as a country?

55:53

Okay, so let's just go through the three major populist arguments. So you touched on them. So the first is AI will destroy all the healthcare jobs.

58:36

Yeah.

58:46

Okay. The second one is AI makes mistakes and patients will die.

58:46

Yeah.

58:51

And then the third is that AI takes away patient choice and autonomy and creates confusion for the physician.

58:51

Right.

58:58

Because it's that can the patient choose or is the doctor required to get the second opinion anytime? So let's just go through each in turn. The AI describes or the. Excuse me, the AI will destroy the healthcare jobs. Yeah. Go against that one.

58:58

I mean, as in you want me to say, like why we shouldn't.

59:15

Yeah, like argue against the populist case.

59:17

I mean, the argument is, I mean, that one is easy, right? Which is like everyone always complains about the administrative cost and bloat of healthcare, so we should definitely want to decrease that. And not even just from a cost perspective, but from like an experience perspective for everyone. No one likes dealing with the administrative part of health care. So it seems very obvious that we should try and remove that and actually maybe that will push more of those people into jobs and roles in the health care system. That don't exist but actually would be really useful like proactively taking care of people or checking in on them or doing more community related stuff for patients in the communities of the providers themselves. So those are like bad, those are like bad tasks and jobs and I think are just like we shouldn't have them. So I don't, I think that one is like the easiest to be like, listen, that one, that one's arguing for admin men.

59:20

Yeah, that one's easier. By the way, that's always been the anti technology argument.

1:00:10

Sure.

1:00:15

You know it's automobiles were taking jobs from horse trainers and ATMs were taking jobs from, from bank tellers.

1:00:16

Yeah.

1:00:22

And you know, in a lot of those cases, not all, but in a lot of those cases we have more of that prior job but just doing something. We have more bank tellers today as an example. Right. So yeah, so. So I think that's one big one. The other one is just that it's super supply constrained. So I think in some other markets there probably will be some more drastic job dislocation. But I think in health care it's just a simple supply demand mismatch. Right. There's just we're employing more than any other sector and yet we still have record shortages. Yeah. And so. And it's not getting any better. Right?

1:00:22

Yeah, yeah. So there's an unlimited demand for health care.

1:00:55

Unlimited demand.

1:00:58

So if you enable people to get more services, they're going to get it. And that's also probably not a bad thing. Right? Yeah. People want to engage with their health. Like we should actually probably let them if they really want to.

1:00:59

Okay. Go through the second. So I makes mistakes and patients will die.

1:01:09

So I think the case here is more around like we have to think about type one and type two errors a little bit. Right. Where it's okay, it's obvious that AI is going to make a mistake at some point and something bad is going to happen. But how many net people did it help? Right. And you have to weigh the costs between those things like for every. And you even see this just like anecdotally on the Internet for every person that it's probably going to make a mistake for. It's also has one person that was like I was never diagnosed and I found out I had this rare disease thanks to dumping all my files into Gemini. And now I learned that I can be saved some random case like that, but also just like the median level of care. If it gets better then it's a little bit of just Risk reward for every mistake that happens, how much benefit is going to happen? And I think the answer probably is in the median. AI implementation at some point is going to raise the bar for everybody, even if it means a few edge case mistakes. And I think you have to be comfortable with that if you want to roll this out at scale.

1:01:12

And humans make mistakes too.

1:02:19

Humans make mistakes too. Again, the Doctronic one I think is just an interesting case study here where it's like actually what you would want to do is compare maybe the Doctronic autonomous AI to you're filling out a form for asynchronous erectile dysfunction company. Like they probably make mistakes too on the other end. Right. But if it's the same level of mistakes, then we should maybe be okay with it if the cost is cheaper, but also maybe it's actually better. Right. So even if it makes some catastrophic.

1:02:20

Errors every so often, there's so many counter arguments here. But the one that interesting, which you already alluded to is that AI creates an audit trail because everything is documented. And so, you know, people are really concerned about the black box of AI, but you can see the decisions at least now in this way as a sort of loop. And so that also creates an interesting data. Like you have data provenance. You can look back and you'd see why a decision was like, where in.

1:02:50

The reasoning did it mess up to actually fix that.

1:03:16

Right.

1:03:18

And not have that happen going forward.

1:03:18

Right.

1:03:20

Imagine trying to deploy that fix at scale with all the human doctors.

1:03:21

Can't do it.

1:03:25

Right. It's really hard. That's actually why we have evidence based guidelines. Right. Like you are trying to prevent a lot of those cases. And in order to do that you have to deploy care guidelines. But if you could make a change to an AI model at scale, actually maybe that's easier.

1:03:25

The third is around AI takes away patient choice. So if your insurance company uses AI for prior authorizations, you're not opting into that.

1:03:40

Yeah.

1:03:50

You know, it can be dehumanizing. You know, now I'm talking to an AI chatbot, but I really want to talk to a human. You know, it's related to the physician piece where now you're second guessing a doc every single time. What do you make of this?

1:03:50

This one is a little harder. But again, I do think this process is already happening. It's just with a human in it, right?

1:04:05

Yeah.

1:04:12

And because of that it can be really. There's a lot of variability in who you talk to on the other side of things. Right. You'll see lots of patients, for example, that, that know which person you want to hit for customer service to get the right answer and blah, blah, blah. Yeah. And there's a lot of variability in that. So maybe this provides more standardization at least. Like you understand the rules of the game and you have to. Basically that's how you play it and it's more understandable and you get answers faster. At the very least, even if it's not the answer you want, at least it's quick. But also at the same time, again, it's just you always have to compare it against the human being, I think. And the core problem I think there, especially with like payers prior auth, all that kind of stuff has nothing to do with technology. Technology. Right. It's really just there's an incentive misalignment between the payer provider and insurance company. In a lot of cases, the AI is probably just like encoding that, but it's not that different than deploying a call center agent with rules. Who needs to follow them. Right. So again, it's the same as the other ones where it's like, compared to the human version of that. Is that better or worse? If you can get answers faster, I would argue it's better. Right. No one wants to be like waiting to hear from the insurance company if the prior auth went through. At least get the decision quickly so you can move on to the next thing at least and argue or whatever it is, but just get the decision quickly.

1:04:12

Yeah, I've heard this argument. Feels like the luxury argument to me.

1:05:34

Yeah.

1:05:37

Where sure, if we could have one doctor for every patient and they would get 247 attention.

1:05:37

Yeah.

1:05:44

Then sure, maybe AI takes away from that, but that's not the reality. The reality is we've got a labor shortage and we've got long wait times. And so if you've got a AI agent that's 247 available infinitely patient, has all your medical context and getting back to you quickly. Yeah, I think people want that definitely. And I think we, I think in areas where we can create choice, I think the data, my expectation would be the data would suggest that people want.

1:05:45

That, but the bot's got to get better, man. Have you like called an AI bought recently? Like, I mean some of them are real bad and there's, I'm like pressing zero to call do it. So the deployment I think actually really matters here. And my worry a little bit is that payers have no incentive to actually make this like a good experience.

1:06:11

Yeah, that's like A version of the Turing Test is at what point do you not automatically just say representative?

1:06:29

Yeah. And they're like, I am a representative. Excuse me, I'd like to not talk to a bot. It's like, sir, how do you know?

1:06:34

How do you know I'm not a representative?

1:06:42

Oh, I'm sorry. I. Did you.

1:06:44

Exactly. Okay, so. So we talked about AI and so. So we think dollars are going to go in different places there. They're going to spend it on screening, they're going to spend it on AI applications. Now let's talk about drugs. So you had a prediction in there. GLP1 usage rates will more than double, but they won't work well on mental and behavioral health issues.

1:06:46

So set this one up. Yeah. So the big thing this year is that the pill version of WeGovy is coming up, right? So, you know, suddenly. And now also there's a ton of negotiations that have happened, both from the government side, to bring these costs down, as well as, you know, pharma going direct to consumer, which is a very new thing. And also the PBM rebates, right? Like, everything, this is such a competitive drug class that the prices are going down. Right. And new modalities now are being created, like the pill form, and you have these, like, new stronger doses or different combos like diadem and red. A true tide and all this kind of stuff. Incredibly competitive drug class, right? So. And it is also, again, like, basically near infinite demand. Like, people. Everyone wants to not be, like, obese, right? Near infinite demand here. So it's clear that a lot of people are going to be taking this drug, right? It's just clearly going to happen. I think I forgot. I think maybe last year we were like 10 to 12%, if I remember correctly, or so. Maybe 15, somewhere like that.

1:07:05

That sounds right.

1:08:09

So now with this pill coming out and the prices coming down and everything, like, it's gonna. It's gonna go up for sure. I mean, the percentage of the US population that is obese is, like, gigantic. So the number of people who want to take this is gonna go up. I don't think anyone's gonna. Gonna doubt that a lot of people are gonna be on this drug, right? It's just. It's gonna be high. My guess was 30%, just based on where we were at a baseline, but. And it's always nice to put numbers around predictions. So let's see if I'm right about that. The mental behavioral health one is interesting because this is, you know, it's funny. It's like in this whole vein of people doing self experimentation, one thing they have been doing is around let's try JLP ones for everything. Right. And a very classic area. Oh, it helps my mental acuity when I like microdose wegovy in the bathroom. Right. So people have been experimenting on this and you know, they're like, oh, it brings me so much mental clarity and lots of theories around how the actual thing that, that these drugs are solving is really a dopamine regulation issue on how you think about food. So really regulating food addiction essentially. Right. And that can that cross over to other addiction types. Right. But a lot of trials in the past have actually in a similar ish mechanism of action have tested some of these areas like in smoking and alcohol and all this kind of stuff. But like not fantastic results actually. So maybe you know, changing the dosages and the formulation is all that kind of stuff does change that, but it doesn't seem again this is just a guess and I'm not a doctor, I'm just like a dude with a newsletter. It doesn't seem like it's going to work that well for addiction. And even you know, they tested in Alzheimer's, it didn't really work out that well. So I think they there's still a lot of room here to like experiment. But my guess is just for mental and behavioral health, it's not gonna work that well. One thing I did actually hear recently, I have quite a few friends slash people who just slide in my Instagram DMs for unclear reasons about their patient journeys. And whenever I post about GLP1s, people will tell me like a different GLP1 story that they have. And a lot of people with like inflammation related diseases so think like Crohn's and ulcerative cortisol colitis or psoriasis for example, have found that microdosing the GLP one actually seems to be helping them a lot. Like one person just recently told me they stopped their biologic and the GLP one seems to work for them. And that actually might be a different interesting area that it might work better than we expect. But I don't know mental behavioral health stuff, I don't think it maybe is going to work as well, but I could be totally wrong about that.

1:08:10

But we're going to learn a lot.

1:10:50

It's clear trials also, I think readouts.

1:10:51

This year it's clear that there's a lot of trials running on a lot of different things. You're the meme of the GLP1 Golden Hanger Hammer. You know, why does it work on everything I touched? I mean, it's. The efficacy is dazzling. It's crazy how many places put in.

1:10:53

The water supply, man.

1:11:07

It might be like the next statin or the next just put in the water supply kind of thing. So, you know, I know less on that part. But I would very much agree that there's basically five accelerators that are just going to. It maybe goes past 30%. I mean, we'll see. I mean, the consumer awareness, I think the dazzling efficacy for a lot of conditions has captured the intention of a lot of people. It's in the zeitgeist. It's out there. One you didn't talk about, but I think is important is just the rate. And this is true for glp. One is true for all new medications, therapeutics, devices, I think, which is just evidence dissemination is also happening faster. Certainly tools like Open Evidence and chatgpt docs using that means the information gets out to the community, treatment patterns change more rapidly. And I think there's some data to support that, but also to consumers.

1:11:08

Right.

1:11:57

Because we're playing around with these tools and they're telling us things. The convenience factor, no question. I think the oral formulation is a big deal. The new channel, relatedly, the new channels I think is something that I expect we'll see a lot more of, which is, you know, you had the oral formulation of WeGovy on day one of launch effectively go live on all these digital health channels. So it was on RO. It was on GoodRx.

1:11:57

Yeah, yeah.

1:12:24

That one's kind of a newer phenomenon.

1:12:25

Yeah.

1:12:27

It used to be that you'd go your mainstream channels that your commercial pharma org totally has you, you know, you know your people, you know, your docs. You're going to talk to the docs that are the influencers, the KOLs. Yeah, you're going to go talk to the docs where you ran the study.

1:12:27

And we're talking the TikTok influencers now. They're the KLL.

1:12:41

Now you're talking straight to TikTok.

1:12:44

Right. It's crazy. Yeah, so.

1:12:46

So I think that is. Is another channel that is only getting bigger and then I think the fifth is just around reimbursement and payment. I think to your point that some of these categories are complex. I think the formulations are changing or our patients are getting more formulation options and those may or, you know, have different prices or different. They may distribute through different channels and that'll have its own. There'll be their Own competition. Then obviously, you know, Trump Rx is going to bring prices down and, you know, you're gonna, you're gonna just have a lot of, a lot more channels and a lot more options. So I'm excited about that.

1:12:47

Drugs are a consumer product. They always have been.

1:13:21

Yeah.

1:13:24

Obesity is just the area which has so much demand and people understand their own disease pathology so well.

1:13:24

Yeah.

1:13:31

That they know how to buy like what they want. Right, right. Versus it's hard for me to be like, hey, I know what cancer drug I need or like rheumatoid arthritis or whatever. But it's, hey, here's a pill that actually helps with obesity. I understand that I have obesity and I, you know, it's the first. I think it's very cool to see a drug company treat itself like a consumer products company. It's very rare.

1:13:32

Yeah. And I think these, I think some of these channels where I think you, if, if you've got now, you know, effectively an AI doctor in your pocket, whether it's physician supervised or not, gives you another channel to have conversations and understand what's right for you and what's not. I think a lot more patients are going directly to their doc and saying, hey, should I be on a GLP1? Right. So there's just a very different, it's a very different level of poll versus like the old school pharma ad that's ask your doctor about X and then let me tell you the 15 side effects.

1:13:55

Yeah, yeah. And they're like having a grand old time in the field and everyone's like, you might shit yourself to death.

1:14:24

Yeah, exactly. Okay, so that's one drug class that's growing very quickly. There's another category of drugs, of course, other peptides. So not GLP ones.

1:14:30

Yeah, yeah.

1:14:40

You have a prediction the FDA will crack down on gray market peptides and create regulations for legit compounding facilities.

1:14:41

Yeah. NASA and I've been trying to find these compounding, these peptide parties and no one's invited me. I'm okay, I'm a little offended.

1:14:47

Okay, so we're going to, we'll show a picture of these peptide parties from that one article. I mean, it's crazy what's happening with right now. So let's set this one up. But you're predicting the FDA cracks down on gray market peptides in 2026. The plot twist is basically. So your prediction, I think, came out at the end of December, six or seven days later, RFK Jr says FDA's war on public health is about to end including its aggressive suppression of peptides. Y. Okay, so for people that don't know. So peptides are amino acid chains. There are some FDA approved peptides, GLP1s are an example. But then there's a whole category of unapproved peptides, many of which are coming from China or at compounding pharmacies. Both are unregulated or gray area, you know, areas and people are trying them. And there's peptide parties, as we've discussed. They look like raves, but everyone's drinking non alcoholic beer and soda water but then injecting themselves with needles. And so it's a.

1:14:53

They're worried about what alcohol do.

1:15:55

Right. So you know, on one hand maybe we're becoming healthier and then maybe we're trying new things.

1:15:57

Seriously. So.

1:16:02

So. Yeah, so. So sorry to interrupt you, but lay that out.

1:16:03

So I mean it's actually interesting because you know, you would think it's maybe just like the small biohacker types, but like I know quite a few people who are older who take it for things like pain, like BPC, I think 157 is a very common one that people will use and inject in their joints to help alleviate pain. And I actually think one of the things that shows is for their areas where there's like unmet need, right? There's not really a lot you can do for joint pain or whatever that people are. Because pain is so bad they are willing to try a lot of things, right? Including these like really experimental drugs. So I mean the prediction is not that we're not going to have more peptides in the U.S. it is that the current way in which we procure peptides, if you want to do them, is through very shady sources. Like to the date people are either buying it online in these sites that look like, you know, infomercials from the 90s, or they're going to their doctor and they're like, hey, where can I get this? And the doctor has some relationship with trusted compounding pharmacies that they have seen or whatever. And it's just like all over the place, right? There's not really any regulations around how we procure source, you know, measure the, for example, like the dosages and all this kind of stuff. It's just like total wild west. So my, my theory is that there's, you know, what I thought was going to happen is there's gonna be some again, big news story or whatever of a bad batch that like really harmed a lot of people. Then suddenly they're like, we gotta do something about this. And then the US would basically create pathways to create more accredited versions of compounding pharmacies effectively. So that's like my guess on what happens, I don't think. And again, this kind of brings it back a little bit to the GLP1 conversation, which is there's another part of this which is like, the question of how does. How do we think about IP laws around drugs in general, right? Where it's like, if I come up with a new drug and I'm like, hey, here is this new drug that cures joint pain, and someone is like, I'm gonna make a compounded version of that and charge one tenth the price. What is even the point of having IP laws in the first place? Right? So I think that brings up a very separate and interesting conversation. But this administration, I think, is more on the side of we should just make things cheaper for everybody. And, and, you know, pharma basically is maybe losing that battle. Sounds super clear to me. Or. And, but. And the other thing with peptides is there's a whole range of them, right? There's peptides like retatrutide, which is actually still being studied in clinical trials today. This is just sort of like a shortcut way to get them faster versus, like, totally unstudied ones that do God knows what. Right. So, you know, if you go to these, like, bodybuilding subreddits and all that, they're always on the fringe of this stuff.

1:16:05

Stuff.

1:18:58

Of course, it's a little bit of survivor bias. The guy who, you know, injected some random peptide is not posting on those forums anymore. Yeah, they're trying all this stuff and question is, should we actually be capturing structured data from these people if they're doing it anyway?

1:18:58

Totally. Well, so first of all, it ends up on a subreddit and then it ends up in an LLM. Yeah. And then you have an LLM that's saying. So you got kind of dangerous loops there.

1:19:12

Exactly.

1:19:20

You know, this is a complex one, because if you ban imports of these peptides, then you're. You probably result in even more underground movement here. And then if you make. If you approve certain compounders as compound pharmacies, as you're allowed to do this, then do you. Are you informally suggesting that you're okay with this and then encouraging it? And so this is complex.

1:19:21

I mean, I think they are sort of doing that. Right. Like, it is the. It all falls under this, I think general ideology around patients should be able to choose and patients should be able to absorb risk themselves in exchange for lower prices if they so choose to do so. Now again, totally depends where you fall on the spectrum of do you think patients should be able to do that or not? Can they make good decisions here?

1:19:51

This is the freedom versus public health argument. And you know, it's as old as time. I think the thing you, you alluded to this, which is it would be a real tragedy if people are going to do it anyways, no matter what happens. And then we don't learn anything from it. So if people are going to do it, can we create what we've done with AI, where we're creating, you know, in certain states are creating sandboxes.

1:20:14

Yeah.

1:20:34

Where we can learn from it.

1:20:34

Yeah.

1:20:36

Can we create that here too?

1:20:37

Yeah.

1:20:39

But the problem is the moment you.

1:20:39

Open up the door a little bit, you're endorsing it. I know, exactly. If you're like, hey listen, I mean, this is basically like the safe injection site version of peptides, right. So as soon as you say, hey listen, send us your data, you're basically saying, go do it.

1:20:40

Yeah.

1:20:53

And I don't know, do you really want that? Are we going to be taking peptides in 10 years?

1:20:54

Really hard?

1:20:59

Can I tell you an interesting.

1:21:00

Please.

1:21:02

So in the US we run three phases of clinical trials, right? Phase one, to taste test for safety and healthy volunteers. Phase two to test for efficacy to see if the drug actually works in small populations with the disease. Phase three to see if that replicates across larger populations in many geographies. Blah, blah, blah. Right. There is a movement out there that is basically, and it's coming to the US now actually, that is basically like why do we have to test for efficacy at all? Why don't we just test for safety and then let the market decide what efficacy is?

1:21:03

Right.

1:21:34

So as soon as a drug is proven safe, then we should just be able to. I should be able to go get it if I want to. Right. And you'll bring drugs to market way faster that way. And then the insurers will basically look at the efficacy data to see how much they'll pay and reimburse it. And then over time you'll have more real world evidence to say, hey, this is working, this is not, but people can make their own choice.

1:21:35

Yeah.

1:21:56

I'm not saying it's right or wrong, but like Montana, I believe, passed a law that allows you to do this. I think for gene and self defense therapies where you can, you know, I think go and try a cell and gene therapy basically if it's proven as just like doing phase one. And they've been doing this in the Honduras, in this, These like charter cities to. For, you know, gene therapies for Follistatin and all this. This is the.

1:21:56

Brian Johnson.

1:22:19

Yeah, yeah, yeah, yes. So that's, I mean, peptides are sort of an extension of that ideology a little bit where you're like, maybe I should just be able to try this.

1:22:19

Myself and see, you know, the libertarian sort of free market, you know, person in me likes that. But I think the challenge is then, okay, what are you allowed to market? So if you've only tested and demonstrated safety, then what are you allowed to say to convince and persuade people to try this thing? And can you make medical claims around efficacy if you haven't tested it?

1:22:29

Yeah, that's a good question.

1:22:53

You know, and then it's a slippery slope.

1:22:53

Yeah.

1:22:55

And then you let that, that you let one person make a claim and then all of a sudden.

1:22:55

Yeah, yeah.

1:22:59

The ones that truly are very efficacious become very difficult to decipher from all the ones that are just saying things.

1:23:00

Yeah. I don't know. I don't know the answer to that. And again, this is like maybe where patient biobanks become interesting, where you're like, can I interrogate a registry of outcomes and see what it did anyway?

1:23:07

Yeah.

1:23:17

But not totally sure what they're gonna do about that.

1:23:17

It feels more like food too. Where, you know that I'm not an expert in food testing, but.

1:23:19

Yeah.

1:23:24

You really just have to make sure it's safe and it's supplements. Yeah. Somewhat edible.

1:23:25

Yeah.

1:23:29

But then you let the market decide if it's efficacious. Is that good? I mean, you know, I mean, everyone.

1:23:29

No one likes the US Food system.

1:23:34

Is it the US Food system good? I don't know. So we're working on fixing that.

1:23:36

Food has that concept of generally regarded as safe. Right. Like that if. As long as that you can rip it.

1:23:39

Yeah.

1:23:45

Actually, maybe we should have swapped the two.

1:23:46

Oh, God.

1:23:48

And actually food should have gone through more testing.

1:23:48

Yeah.

1:23:52

Before it comes to the system. Because it's at scale.

1:23:52

Yeah.

1:23:55

For everybody.

1:23:55

Yeah.

1:23:56

And then maybe drugs should have been generally regarded as safe. Yeah. Maybe that's the way we should have gotten.

1:23:56

Okay, so we covered some of your more spicy predictions. There's more in there that I thought were interesting. We don't have time for them. Check them out if you're watching. We'll link in the comments here. But one you were sort of predicting out of Pocket. So tell you know what's going to happen out of pocket and make a little prediction there.

1:24:02

We're like all in on events now and in person and irl and I think maybe this is like a broader content thing that's happening. But just I'm just like not having fun on the Internet anymore. I feel like AI has like really just distorted how we, how content is created and then the algorithms are sort of boosting stuff that like doesn't bring me enjoyment anymore.

1:24:19

The AI slop is killing you.

1:24:39

AI slop is killing me. And like the, you know, when posts are like, like becoming a part of the zeitgeist. Like it's just now I'm getting mobbed rather than like having fun. Like the discussions are not as interesting to me and I'm not learning nearly as much online. And so we are all in on events now and so we're doing hackathons, we're doing, we have this like micro conference format that's like all workshop oriented, really role specific and also like, you know, we're actually doing a lot of like workshops teaching people like interesting ways they can use AI in their work. So we're like doing a lot more in person stuff. And then also like for me personally it's kind of interesting running like a creator focused business or something because it's hard to scale. Right? It is like an interesting question of like how do you scale a single person. We have been exploring a lot of AI tools ourselves internally. I've been clawed code, just grinding, you know, and how do we use that in different parts of the business, like scale up ourselves basically. But yeah, really it's like a big focus on events this year and just, just doing more fun healthcare stuff. Keeping healthcare fun, you know, keeping healthcare fun.

1:24:41

2026. I mean even since you wrote the article, just we, you know, we were at JP Morgan this week, the news is nuts. Things are happening at crazy pace. So at least in probably in our careers, I'm sure you'd agree this is the craziest time in healthcare, period.

1:25:42

I will say it is the first time. I mean I've only been doing healthcare stuff for a decade so like take that with a grant, sure. But it's the first time. It sort of seems like I see a path to either deflationary or flat spend in healthcare combo of AI, GLP1s, you know, an administration that's kind of trying to focus on bringing costs more in line. It's just an ideology in the form of like consumer spend and all this. Kind of stuff. Like I do see an interesting path now to maybe getting spend more in control. Tbd. I mean the timelines on that horizon are like very unclear. But there's just so much cool stuff happening.

1:25:56

I'm equally as optimistic, but I see it differently on what happens.

1:26:35

Okay.

1:26:39

Which is that I think healthcare is going to get a lot, it's going to get a lot better, but we're going to spend even more on health. And it actually has nothing to do with what's happening in healthcare and it's more just a function of what's happening to the country which is as you know, people thought that healthcare was going to bankruptcy bankrupt us for a long time but we keep spending more and more on health care because our GDP keeps going up and we have more other things are getting a lot cheaper. Electronics, you know, basically a lot of the industries that software has eaten have gotten a lot cheaper and it allows a country to spend more of its money on other things that it values. Yeah, healthcare, education, etc. And by the way, you know, real estate and a lot of those countries, a lot of those sectors have become a lot more expensive. You know, five, ten years from now are we going to be spending more on health care? I think we're going to be spending way more, but we're just going to get a lot more for it because these things are going to get better. We'll see. I think either way it's a really fun time to be.

1:26:39

Let me do a little further out prediction sort of related to this which is as the US moved from a manufacturing based economy where that went overseas, we became a services economy. Yeah. A big part of that trade was that healthcare basically became a jobs program to effectively replace manufacturing. Right now if we really want to see the productivity gains from AI, we have to figure out what the next thing is. Post services economy.

1:27:38

Right.

1:28:10

And that's not just healthcare, that's for a lot of stuff. Right. Like you know, a lot of the jobs that are white collar jobs are all we're already seeing as like sort of aiable. Right. So we need to figure out something that's like a little bit post services. One theory I have is that we are going to move to something that looks, it's hard for. I don't know exactly what the term is but maybe it's like a community economy kind of thing.

1:28:10

Creativity economy.

1:28:32

Creativity or just you know, help your fellow person economy. And healthcare actually provides the existing rails to already do that. Yeah. And one form of that for example is like Paid caregivers. Right. Or I mean, what effectively almost looks like paid friends, essentially. Right. Like you see this in Medicare Advantage. Well, they'll pay people to like to come just take care of some of the older folks, et cetera. So actually that might be the way that, you know, healthcare spend sort of absorbs a lot of this job loss is just by basically saying, hey, we think that the way that, you know, the next economy going forward is like people just feeling good about their lives and not necessarily just from a pure healthcare perspective, but health care provides the rails to do that essentially. And that's why a lot of states have these paid caregiving programs, for example, that are some of the fastest growing jobs in those states.

1:28:33

Yeah.

1:29:27

So, you know, that might be one version of like how health care spend increases.

1:29:28

Yeah. I think the jobs thing and then what we spend are sort of two separate but very related things, like maybe we have fewer healthcare jobs in the way we define it today as a percentage of overall, of the overall population and where people are working. But if, if it's something that we value more and there's new exciting stuff and we can afford it. Yeah, prices will rise. But we'll see. We'll see.

1:29:32

Yeah, yeah. I don't know, we'll. We'll still be in this industry or that happens, that's for sure.

1:29:57

No doubt.

1:30:02

And we'll be still faxing. Yeah, exactly.

1:30:03

We'll be, we'll fax it. We should have just done this podcast by fax.

1:30:05

I know. It would have been more fitting.

1:30:07

That would have been more fitting. A lot of people, you know, what I've seen, which I've been really excited about too, is technology broadly. And people who are working in technology, there's an increasing percentage, this is anecdotal, increasing percentage of people in technology that are coming into healthcare.

1:30:09

Sure.

1:30:24

They're interested. You know, I think stuff that you're doing to educate people about the complexity of healthcare, it's not actually that complicated. Come learn it, come check it out. A lot more people are entering and I think that's awesome. I think that's why, you know, article, that's why I wanted to feature this predictions piece as an example of just, you know, there's cool stuff happening in healthcare. Come learn about it, come make it, come make an impact. But in the age of AI slop, it's hard to figure out what to read and what to work on. So what is your content? Diet?

1:30:25

Ooh. One thing that I've been doing recently is I'm going way Back to primary source stuff now, partially because the AI tools also now make it easier to parse through, like, really long documents and just really focused on the parts that I want to focus on. But, you know, like, my most like, old head behavior now is like, I print out like a lawsuit and read it in the morning or I get print articles or long form pieces or white papers and stuff like, I'll print out and actually read it. Yeah. I think there is, like, actively more wrong information on the Internet now than right information. Like, I think that actually has tipped.

1:30:54

Yeah.

1:31:30

And so now it's time to go back to just reading the primary. And also I just am having more conversations with people directly where I'm just like, what are you experiencing? What's interesting about what you're doing? And I think just talking to people about the stuff they're working on and then following up on some of the things that they talk about is just way more fruitful and interesting. Now at this point, versus trying to read online and believe that is truth, I've now crossed that Rubicon. So I read a lot of white papers and stuff like that, and then there are a lot of substack people and a lot of people who I think are doing a good job of writing, like, what I think is more sort of reliable information and learn a lot from them.

1:31:30

Yeah. I found LLMs have been a great translator of complex scientific white papers.

1:32:12

Yeah.

1:32:18

That if I'm not an expert in that domain, would just take me forever to parse through. And I just. Just drop that into ChatGPT or Claude or Rock or whatever and let it translate for me. And then I can always talk to people about it and go and follow up. But the pace of stuff that is emerging right now, you can't keep up.

1:32:19

No.

1:32:37

You need an interpretation layer to our earlier conversation. So that's been.

1:32:38

Yeah, exactly. Especially like, if you. If you know a lot of the basics and you really just want to go to the more advanced stuff, like just remove all the basic stuff and just show me the stuff that's particularly interesting. So it's been fun.

1:32:41

It's awesome. Well, my man, I appreciate you coming on.

1:32:54

Yeah. For people who don't know, by the way, I was like, yo, Jay, let's catch up. And Jay was like, yeah, man, let's do a podcast. And so that's like, very class, you know, peak male relationship.

1:32:57

This is the modern male relationship is if it's not recorded, why are we even doing it?

1:33:06

Next time we'll have a peptide on the table and we could just take it together.

1:33:10

Yeah, and we just show that on video, too.

1:33:12

Yeah, man. Thanks for having me.

1:33:16

Cool.

1:33:17

Appreciate it.

1:33:17

Thanks for coming.

1:33:18

Thanks for listening to the A16Z podcast. If you enjoyed the episode, let us know by leaving a review@ratethispodcast.com a16z. We've got more great conversations coming your way. See you next time. As a reminder, the content here is for informational purposes only, should not be taken as legal, business, tax, or investment advice, or be used to evaluate any investment or security, and is not directed at any investors or potential investors in any A16Z fund. Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast. For more details, including a link to our investments, please see a16z.com disclosures.

1:33:22