Welcome back to the Stanford Healthcare

AI podcast and we're so excited to be

joined uh by Andy Slavit who needs no

introduction uh but currently is the

co-founder and general partner at Town

Hall Ventures uh and formerly you may

have known him before as a senior

adviser the Biden administration

overseeing the pandemic response and

then ran and CMS under Obama. So,

welcome Andy.

Great to be here. Justin, Matt,

good to see you.

Well, uh, we got to start in the same

place we've started a number of other

episodes with like what is the current

state of the world on AI uh, adoption

and things are still racing ahead. So,

the the data we'll pull up, Matt, for

you here is kind of the current LLM

leaderboards. uh it's been uh no slowing

down in kind of current models kind of

overseeing others but Matt Matt what do

you make of this? Yeah, I mean of course

as we always try to caveat this as of

this recording um uh we see this kind of

very heterogeneous picture I think

that's emerging uh the closed labs are

are still in the lead on a lot of these

leaderboards and again we could choose

from a variety of the different

benchmarks uh but some of these generic

ones looking at text this is LLM arena

where they kind of pit the models

against each other and uh sort of judge

the the broader output I I think that

the takeaways is especially after the

Gemini 3 Nano Banana launch is that

there's a significant kind of pulling

away by the Google models in the image

generation uh category. I think that um

I think that's probably not a huge

surprise. Uh right, they they sit on

some pretty substantial assets including

of course uh YouTube. But I think on the

text and then on the coding side um

despite again the the big leap I think

that Gemini uh showed in both of those

categories, there's still I think a

developer preference for the clawed uh

models and that that doesn't seem to be

changing a ton at least on the on the

full enterprise side. And then as we

look I think everyone's starting to form

I don't want to say relationships that's

a little strong but I think people are

starting to feel as though the apps that

they're comfortable with whether it's

Chad GBT whether it's Claude whether

it's Gemini I think they're starting to

pick their models um and and that seems

to be the trend so I I don't necessarily

see a big leap in the tech side

necessarily changing the game until uh

maybe there's another larger

breakthrough. What I also should flag is

that I mean as a user as you sort of

talk to more folks in the industry

um you know the idea of memory is still

a bit of a it's a bit of a nent product

shape right like how much of the prior

conversations do you want incorporated

into the current convers that's a

subject of debate right but that does

change depending on your use case how

much you enjoy or get value out of the

model's usage right so like I can

imagine a future state where it knows

the things that I care about, maybe it

knows my background, etc. And and and

again, I get better outputs for the

longer relationships I have with some of

these apps. But, um, but then there's

also places where that's not an

advantage, right? Where you you don't

necessarily want it to pull in the fact

that I'm a whatever physician or an AI

researcher. I I actually want to ask it

a question to not include those things

about. So, it's still I I still feel

like that's also uneven, but it's still

anyone's game. And I I think that the

one of the biggest takeaways for for all

of you in the audience who maybe don't

follow this as obsessively as we do is

um that there's no slowing down. I think

there was a couple points over the last

you know again remember this technology

is essentially really only been in the

public conscious say for three years

which it it feels a lot longer than that

right but um but I think that that it's

it's still early enough and the

acceleration is continuing. So, I think

any narratives that I I maybe last

summer that came up around hitting a

wall or slowing down, I think that um

has has at least for for now not uh not

proven true. So,

and andy, we were just chatting a little

bit before we hit hit record. You know,

you said you guys were using, you know,

you saw Google on the top of the charts.

You said you guys were big Gemini users.

uh where where are you seeing your own

usage you know today where you're

starting to use AI in your daily life or

your work where else are you seeing it

my sense is that our team for individual

individually have the preferences and I

think it's probably been open AI from

what I've been able to see chat GPT

um but at work we chose to to use Gemini

in large part because we're a sorry for

saying this we're a Google shop and um

uh you know and it did it just the

integration into searching your emails

and your Google Drive and everything

else um just uh was easier. So we're

we're like torturing our team by making

them use something that they that they

don't prefer and then I have feeling

they keep you know two windows open on

their two screens, one to one one for

their personal stuff and one for their

work stuff and um but we'll you know

we'll change. I mean, I think we we are

very open to um keeping up and adopting.

Um we're we're about to bring in an AI

research fellow full-time on the team

whose job is going to be solely to keep

us um at the cutting edge. And so I have

a feeling he'll come in and start as

soon as he's done laughing at where we

are um he'll he'll make some changes or

she'll make some changes.

Well, say say more about that, you know,

like where where did that come from?

When did you decide, you know, it's an

interesting thing, you know, for other,

you know, healthcare organizations

listening or thinking about this, well,

like when did you decide, hey, wait a

second, we need to bring in someone to

kind of get us into shape. You go

through this thing as an investor

where the first thing you're like is, oh

my god, it's hard to invest in

healthcare AI. Um, it's hard to pick the

winners. it's really hard to find the

great companies and then you go to wait

a minute your next discovery is

shouldn't our current investments be

using AI a lot more than they are and

then you like you go down that rabbit

hole and and then eventually you get to

wait a minute shouldn't we be using AI

more if we're going to tell other people

to use it more and I we you know we we

went on the journey in exactly that

order uh and and so um we actually um

realized wait a minute we can't actually

sound smart at something we read about.

We actually have to get much much much

better at it. And you know, we have um

young younger team that I think is very

comfortable and very native. Um, and

then we've got us older folks who are

excited about it. Um, and are, by the

way, it gives us a great opportunity to

learn from our younger team, which is

healthy on so many dimensions. Um, that

they get to be the teachers and we get

to um, really learn from them. And we've

just said, hey, get get us on a core,

set us on an agenda, like what should we

do? And, you know, they've helped us

adopt, you know, a bunch of tools like

granola and and and other things. And

and then this idea for this research

fellow

uh was actually uh one of my partners,

David, uh said we should do we should um

really go um whole hog in this and I

think it will make us better investors.

I think it will make us better working

with our companies.

It it's just this such an interesting

journey right now and I know Matt and I

have you know both been asked many times

hey come to a healthcare organization

give kind of a keynote on AI like what

should we do what are a few takeaways

and one of the things that I you

mentioned that I like to push the

leaders on this kind of joke like AI

can't be delegated

and what do I mean by that it's it is

moving too fast to say oh we'll make you

know, some AI bet and I don't have to

think about this as a leader of the

organization and then we'll and we'll

we'll pull up some charts here in a

minute too just to talk about the

numbers and the change, but things are

moving too fast where as a leader if

you're not it's not that you need to

spend every day or be as obsessed as

Matt and I are about staying on the

current top models, but if you're not

continually checking in for what's going

on, what's possible, you're going to

miss, you know, where you need to take

your your organization and and what's

going on because thinking about, oh, I

tried tried chat GBT two years ago. It

hallucinated didn't do what I wanted it

to do. So like, oh, AI is overhyped.

It's like, well, models are a different

place today. Capabilities are different

and there's some investment and you

know, leader has to make that choice for

how much to kind of stay on top of

currently that's happening.

Yeah. One of the things I've heard you

talk about, Justin, is just how much of

this is is really has to get into the

culture and how much of it actually

seeps into the culture whether you plan

it or not. And and uh so you you can

sort of choose how you respond to that.

And I think that's really good insight.

I mean, I've always found that, you

know, good technology can't be pushed on

you or it's not good technology. And I

think that's what we've had in health

care with you with electronic medical

records and um you know personal health

records everything is like let's change

people's behavior let's push them to

adopt this new technology well for the

first time you know in AI and I think

you've got charts that are that are

showing this people are saying no we're

pulling this towards ourselves because

we want to use it because it's making

our lives better it's making our jobs

better and and I dare you to keep up and

I think very exciting place to be

I I I think that's it. I mean, and I

think if we dial back the the timeline

to, you know, some of the earliest

quoteunquote AI healthcare, it was still

kind of a softwareish thing like it was

not something that the leadership felt

like they needed to use in order to make

a decision and sort of deploy it. To

your point, very much the change

management, very much a we need to get

ROI out of this software. Now it but but

now I almost it's it's it's definitely a

bottoms up like you have you've seen it

in your you know your or clearly

healthcare is seeing this. I mean

Justin's got some data probably worth

pulling up just to show this but it's

almost as if like at home I'm using like

email and I go to work and I'm sending

I'm writing letters by hand like it it

almost has that feel especially for

those who are really using it a lot in

their personal lives and on on their

phones. And so I I think this work this

is from Graham Walker's um off call

survey. So, you know, it it's probably

not a massively representative sample,

but it's a pretty good look at what we

see too, I think, in our day-to-day. And

this,

right, I mean, it's incredible to see

that. and maybe even under reportported

to some extent. But if you're getting to

the 70% number of physicians using this,

but still that almost higher percentage

is saying it's making them better at

their jobs, that may even be despite

that their institution or their

organization hasn't really deployed

something. So that's kind of begs the

question, what are they what are they

using? And you can kind of see that data

start to the frustration level bubble up

because they're seeing the opportunity.

maybe they're using it on their phones

most likely and then you know they they

don't have any some any voice from the

leadership telling them uh we've rolled

this out or we've made this available or

whatever those things are um and then

the lack of influence which I think this

is probably goes beyond just for AI

right I think clinicians chronically

feel as though we are increasingly

uh we have responsibility without

authority I think that's kind of been

the name of the game in healthcare for

some time but um I I think that the

communication, the sophistication level

is clearly being outpaced by the

personal usage and the sort of bottoms

up uh phenomenon. And I think it gets

more challenging as we think about all

of the other things, security,

compliance, privacy, all the other

things that that start to become a

problem. Um, I don't know how you guys

look at these at these data, but like to

me, um, this is about as much of a red

alert if I'm a executive at a health

system, um, that, you know, there needs

to be a way to figure this out and to

empower the clinicians to be able to use

these technologies.

Well, I'll just I'll just call out maybe

just a couple numbers for anyone

listening, but you know what? one one

chart showing 67% daily AI use despite

81% organizational dissatisfaction. So

there's this weird dichotomy of I'm

using this all the time yet I'm pretty

frustrated with my organization and

employer. And on the previous slides we

talked about you know most 71% have no

influence over the tools that they use.

Uh 48% rate their employer communication

about AI as poor. You know zero or two

out of five. And so there's this weird

disconnect, you know, between that daily

use and kind of what people are seeing

and feeling at work. Sorry, Andy, go

ahead.

Yeah, I mean, this is just screaming at

you. This data just screams at you. It

just basically says um this is a

stallion. Uh and and you need to figure

out how to how to how to ride it, not

how to tame it. Um and I mean the for

for me the there's a lot of so what's on

the slides but the biggest one is um you

know you can imagine five years ago

saying to a physician technology is

actually going to make you do your job

better um you know physicians who

suffered very few fools um and who have

suffered through many many terrible

technology implementations

um would would have laughed in your face

and so it just emerged out of nowhere

and I I think it's hard for all the rest

of us to kind of change our frame from

how do we protect people from technology

to how do we let them um run.

That's that's just a different mindset

that I think you're pointing to that um

people got to come to grips with.

There's the extreme case too. I think

there was a there was an opinion piece

or a Substack that I came across and I

apologize to the author because I don't

remember where this came maybe one of

you remembers this but the tagline was

essentially we've made healthcare AI

illegal in this country like that was

the that was the sort of base position

and yes that's very extreme and I you

know I'm not here to debate the nuances

of it but just to say like that is the

feeling I think that some of this data

tells us is that it's essentially being

policed at extreme level almost to the

point where the autonomy of the

physician who can see and feel with

their own eyes and ears and their own

judgment knows that yes, I'm I'm a

responsible adult. I'm a physician. I'm

taking but I know that this is going to

help me in my job. Um the idea of it

being illegal really starts to wonder is

there is there a policy lever here that

starts to make some of this more make

this data kind of even out a bit more?

Does it does it sort of unshackle some

of the the health systems who feel like

maybe that's the barrier. I I I actually

don't know yet whether that's going to

have a big enough influence on I mean

Andy, you're the expert on the policy

side, but I'm just curious like what's

your perspective on some of this because

is there a way to I don't know to make

this so that it's not quote unquote

illegal. I mean, when you've got a

product that's good at open evidence

as people seem to like it. I'm I'm not a

physician, so I I say this based on some

of the data you've shown and my own

experience. Um, good luck making that

illegal whether you're a regulator or an

employer. Um, you know, I I think it's

too late. Um, and I think it's a fool's

journey to try. I I have a lot of

empathy for regulators. How do you

regulate this? So, I'm the only thing

worse than being an AI investor is being

an AI regulator, I guess. Um, because

it's like, you know, you are so far out

of the picture um and the picture is

evolving so fast that um you know to

qualify to be able to render the

judgments to regulate um in this

environment is is really really hard and

I think I have sympathy for the people

that believe it needs some regulation in

order um you know to to prevent the

worst from happening but I have a lot of

sympathy for the people who say how and

how can you do that and and and and why

and it's a fool's errand to try um

government government does do a great

job of regulating things that are steady

state uh let alone let alone moving this

this quickly so to some degree um I

would caution a little bit of patience

um let you know let the let the market

bloom um there will be things over time

that will be um you know perhaps worth

regulating and we can we can address

those things at that time. I don't think

it's a forever thing. Um but um but but

right now I think it's this is much more

about the human um condition and um you

know say I think Justin you're much more

of an expert on this and and you are too

bad is um you know there's a lot of

things that make uh that perhaps are

inhibitors around data privacy and how

all of that works. And you know, perhaps

there's if you want to talk about how to

make people feel safer using AI and it

not being illegal, perhaps there's um

ways to think about how to address some

of those things that that uh that are

that are barriers. But in general, I

just I just caution against thinking

that regulation is going to be a tool

that it will help do anything other than

u make people who render it just sort of

sound irrelevant.

Well, we'll say say more as you know

obviously on the AI side you know Matt

and I spend all of our time but you know

you've worked with past few

administrations and I are in discussions

with folks in the current administration

on all things in healthcare policy but

AI definitely being one of them what's

kind of just your thumb of kind of the

current position for kind of where we're

headed uh again focus on you know AI

regul but also just kind of the other

pieces like what is kind of the current

state for where for where we're going.

The most telling thing I think and I and

I I I wonder how many of the audiences

um have saw the announcement over the

last few weeks over the access model um

which is something that um I um in

iteration earlier iterations I've been

working on with Dr. Oz for about a year.

U you know Oz came to me a year ago as

just as he was starting the job and said

you know if you were administrator

um would you be massively encouraging

the use of AI among Medicare and

Medicaid beneficiaries?

And you know, the answer is um if if you

have a health care system that is barely

touching the people that most need the

care and you have um an opportunity to

really reach people um and keep them

healthier in newer ways, if the techn is

ready, go for it. And I was jealous

because when I was in at CMS, you know,

I was I was putting the brakes on

meaningful use and saying, "Whoa, stop

the craziness, but we were trying to get

people to like actually, you know, use

their computers and I was just like,

wait a minute, this is not the way to do

it." Um, and now we're not we we but but

we knew that if we get technology

distributed, we would have a chance when

really good technology came down the

road. And this is it. I mean, this is

it. This is the opportunity. And so what

they did with this new model um is they

basically said we are going to for the

first time we're not going to just pay

physicians to take care of people. We're

going to pay technologists and

technology companies if the if they can

demonstrate that the technology they're

distributing is making a difference in

keeping people healthier. And it's a

really bold bet. So I point to the

excess model not because I think it's

the greatest thing since sliced bread. I

point to it because it's a signal and

policy makers the best thing they can do

is send signals as to what their intent

is and if you want to read that signal

that signal is they are and the FDA by

the way has created a separate process

um so that things can get approved

without having to go through long safety

and efficacy studies which is really

unusual for the FDA. This is the

strongest signal you can possibly

imagine to say that we want to push AI

through healthcare and do it as rapidly

as possible and reward those who make a

difference and um and I think it's a

really bold move and I think applaud the

administration for it. I think it's more

important than looking just at the

regulatory picture which of course you

know we we all saw the President Trump

signed an executive order saying you

know is going to oppose and try to

counter counter man anything that

happens at a state level on on AI

regulation and you know that's that's

that's their position but it's one thing

to prevent regulations it's another

thing to actively promote with the

dollars and cents of US Treasury as we

are

I and this Annie I'm so glad you brought

that up because you know we've talked

about this many times on our show as

well which is like and and again it

often gets sort of forgotten in some of

these conversations is is that not only

is this technology cheaper, more

performant, and more available to

clinicians in the health care system,

but it's also the same exact technology

for patients. And and and to your point

like you know the and I I think we have

some data that just came out on this too

just kind of looking at how consumers

you know quote unquote which you know we

can call patients at this point uh are

using these tools and the most sustained

conversations interactions are tagged as

healthcare related and that is I I think

those are the green shoots of something

really powerful here because if if

you've ever used the technology and put

in your own data we've talked about this

before on our show, how we do it a lot.

Uh even as you know, board certified

clinicians, uh I still find it to be

incredibly helpful for taking care of my

own health and to help me prepare for a

specialist visit or whatever it is. And

I and I think that that is shared by the

vast majority of patients out there who

are now showing up to their their

offices hopefully more informed.

Hopefully maybe there's a couple, you

know, mythbusting things that still need

to go on the office, but not to the

extent that it used to be where it was

like, you know, you show up with a

Google search and you're like, "Okay,

we're going to spend 20 minutes

unwinding some of this." It's actually

not been that same phenomenon. Now, it's

more, "Wow, you're showing up with you

you've looked at your own health data.

You've looked at maybe it's your blood

sugar data with the model. You've had

long conversations with it, and now

you're coming to me with very pointed

specific questions. I think it's I think

it's an unbelievable opportunity

honestly to start to achieve some of

those disparities that we often talk

about in the show the information

asymmetry that that is in itself a

barrier outside of resources for for

patients to be able to maybe take their

healthcare in their own hands. Anyway, I

I can't say enough good about it. I I

know that there's going to be some, as

we've said, New York Times articles that

are, you know, edge cases and we've seen

some of this with mental health and I

don't want to to minimize that. But on

balance, I just feel like some of these

capabilities in the hands of patients

and thinking about how can I improve my

health and have better relationships

with my health systems. I just see the

net positive here.

Yeah. Well, it's that positive for the

three of us and the three of us have

presumably access to great care anyway.

Um, imagine that Matt, you lived in um

rural, I don't know, western Montana or

North Carolina and it was a nineweek

waiting list to see a physician and an

even longer waiting list to get to see a

specialist. And by the time you showed

up at your at the physician, you weren't

even sure what they knew uh or didn't

know. and they certainly didn't know a

lot about you because you were just

grateful to get an appointment. In that

particular case, it's not just um cool,

it's not just better, but it's it's a

transformative experience with the

health care system and it's necessary.

And one of the things that I I think

we're trying to push at town hall is is

let's create some purpose behind all

this great technology. um let's create

some purpose um that really speaks to

people and you know our favorite

entrepreneurs are those that that want

to solve a really thorny problem for a

set of population that just doesn't have

great access to care and and it's and

it's amazing as you said Matt I mean I I

in an afternoon in an afternoon you can

create your own personal health LLM with

everything in it every lab you've ever

taken everything you've got access to

just U very very simply in any one of

these in any one of these models and

query it any time of day any time of

night. Um and that is that is almost

better than a text relationship with

your physician which is hard to which is

hard to come by.

Yeah. And I I love that explanation of

like we all do it and we can text each

other as physicians. we can find the

very expert, but despite that, we're

still using these tools for ourselves.

Um, and I think it's those two forces,

this consumer lens that we've talked

about again and again in just some of

the numbers we've mentioned before. In a

year, it went from 11% comfort with an

AI doctor to 28% comfort. Again, this is

a Bane survey, small sample. I don't

think those numbers sound exactly right,

but directionally that's where we're

going with consumers feeling empowered

to use these tools to take care of their

health, just like we do, by the way. And

sometimes there's this weird disconnect

of, oh, a patient is different. Like,

no, we're all patients and yes, you

know, we've

as a group know more about healthcare

and how the system works than most, but

we're still patients kind of seeking

that same information and we're doing

it. And that lever combined with kind of

the signaling you're mentioning, you

know, from the administration that this

is the direction,

it's just one of the most unique times

in healthcare where we have multiple

forces just pushing and racing forward.

Um, and on the other side, just because

with the lever we haven't mentioned too

much yet is just the enterprise

adoption, but there's a few pieces of

data that just came out and this was a a

Menllo Ventures uh report where, you

know, they're showing, hey, on the

enterprise side, too, this is the

fastest growing uh growing technology

we've ever seen. You know, people talk

about and you know, I think we've seen

criticism in the comments before, hey,

Matt and Justin, you guys are too

optimistic about AI. Fine. That's that's

probably true and you know people talk

about AI bubbles and things like that

but but there's real adoption happening

at the enterprise side. Uh I I'll skip

this build versus buy but the other

piece that is fascinating now is

healthcare

uh is one of the verticals at least that

they call out as kind of being the

fastest adopted more than legal other

areas uh happening now. And so it's just

this very unique moment for healthcare

where as I have conversations with

healthcare leaders there's also a bit

what feels like a skip the landline

moment where you know healthcare has

been the industry with negative

productivity growth you know uh versus

others because of kind of how hard it

has been to adopt technology and maybe

it's because the technology wasn't good

enough yet but now it just feels like

there's an explosion of adoption.

Yeah.

It's it's it's it's it's incredible. I

look

I'm old enough and I don't know what you

guys are how old you were during the

during the dot bubble, but the closer

you got to things at the dotcom bubble,

the more it felt unsustainable. I don't

know that that's the case here. I think

it feels, you guys are obviously much

closer than I am, but I don't think it

feels that way here. I think it feels

like the closer you get to it, the more

you see the substance and potential. Um,

may be bubble elements um um to anything

that's got this level of of

transformation potential to it. The hype

comes along with it. I mean um how

quickly has open evidence been doubling

its valuation, right? Um that that's you

know that that that's going to happen.

But um at the because there's only a

small number of investors that can get a

piece of that, right? But at the same

time, um, when you get underneath it and

you look at the potential for patients

and physicians and what they're they're

saying, some of the data you shared with

us today, um, you know, it tells you

that that it's it's got a life um, of

its own. Um, that's going to I think

it's going to run no matter what

happens.

I think and I and I I have to agree with

you. I think it's going to I think it's

going to be an interesting time for sure

because I think that the the bubble the

bubbly aspect you know there's plenty of

ink spilled on that on that

conversation. Yes, it involves capex and

infrastructure and all kinds of the bets

that are being made. At the same time

when you when you push that into

healthcare the adoption as Justin showed

massively. Now the question is

uh does the traditional SAS

incumbents can they harness this

technology understanding their users if

fast enough to start to meet the needs

of what is essentially now a bottoms up

as we've said a consumer-based

adoption rate and I and I actually don't

know we've seen plenty of really

powerful use cases obviously coming out

of the ambient space being the key one

the the epics of the world, the Cerners

of the world who are building out in

inside of their UI, inside of their

incumbent space some of these

capabilities, recognizing those pain

points. Is it a fast follow that they

ultimately start to,

you know, meet the needs or not? I I I

don't know yet. But it does seem like

that is one of the biggest questions we

all have, right? Is is it build versus

buy? Do I have to go source it somewhere

else? uh or can I you know sort of count

on my typical technology partner that

I've already invested a lot with to

deliver the needs and the you know the

use cases that I that I want and my

clinicians want. I don't know I I and

again that sort of still strangely

leaves out the consumer patient side of

things I think to some extent although

we're seeing some interesting app

applications through like the my health

announcements and things my chart. Um

anyway it it just seems very early to

me.

Um, despite all of the charts and the

and the fantastic numbers, it still

feels early.

Well, as somebody who looks at a lot of

business plans, um, you know, judging by

judging by these business plans, every

application product use case is is soon

to be a quote unquote wedge into a quote

unquote platform that's going to quote

unquote transform aization.

And and you know, you see that um, uh,

every every business plan has that kind

of as its uh, you know, and then they're

and then they're going to somehow build

a build build a quoteunquote moat. Um,

so it's it's um it's a really funny time

um to look at it from an investor lens

um where you know you're looking for

like what's the and occasionally you'll

run across the true substantial

um u you know capability and you'll just

hope to God that one of the three

venture capital firms that's paying

unlimited prices for for these are not

not yet founded and that you've got some

advantage to them. And I think, you

know, that's where you got I I want to

one other thing you you said um that

which struck me, which is the the the

risk of our friends at the New York

Times, my friends at the New York Times

writing the edge case story about about

the bad situation.

Um

I want to see the stories as well about

the people whose lives were saved

because their very rare cancer diagnosis

um was misdiagnosed by a by a physician

and uh and I happen to know several

cases like this and you we all probably

do and um and AI just basically found

them the one person like them um that

had very similar set of um symptoms and

had a successful treatment and I suspect

that that's happening mult multiple

times every day. Um, and I think it's

happening multiple times every day from

the consumer patient perspective. And I

suspect it's happening even more

frequently when people are using the

clinical tools like open evidence and

others where all of a sudden, you know,

someone who walks into that physician's

office who, you know, hasn't seen, you

know, the latest research paper and

isn't going to spend 15 minutes as

you've talked about, Matt, to to

research it gets that gets that answer.

And uh, I know it's more interesting to

the story about how Whimo, you know, got

stuck at a traffic light because a

traffic cone got got put there. But it's

uh but it's

really important that the stories be

told on the other side of this. I'd like

to think. Agreed. Yeah. And I know we're

I know we're running out of time. Maybe

just one last quick topic which maybe we

can't drain here but one of the things I

wanted to point out and this is the big

unknown. Uh the valuation of open

evidence to your point doubling in six

months I think it's is it 12 billion

something in that range is sort of on

the back of

that's when this episode comes out.

Yes. Yeah. Exactly. But it's on the back

of ads. Okay. And this is the biggest

like uh standoff right now, which is

when do you start to turn on that

infinite money printing machine of ad

revenue and what does that do to your

user base and their choice? And even

again, we can't probably get into it

here, but just to flag it, we know that

these models are incredibly good at

convincing people to do things, right?

We've seen those experiments where they

put the model in a Reddit chat, convince

me, right? and then they have a debate

and they don't know it's a model if you

remember that paper that mildly

unethical experiment I think that was

done but the models won out quite often

and it starts to raise some other very

maybe uncomfortable questions about we

have to pay for this technology somehow

and we also have to be very careful how

we're approaching the ad space and how

we're messaging that ad to again a

population may start to rely on this as

their health counselor, their adviser,

right? And and expect the unbiased

non-ADs version. I don't know. Again,

I'm just raising it. I I don't have a

hot take yet on this, but

I I think it was really smart of them to

make it free. Um and and to make it the

sort of a a deacto standard or at least

for the moment a deacto standard. Um,

but they did kick the can down the road

on the model a little bit. And

healthcare is a little trickier um to do

the kind of advertising that it is in

other industries because of Stark laws,

right? Because of anti-referral laws and

kickback laws and and so forth. So the

natural advertisers

um for for this you know at least

according to current regulations and

there's a place where regulations you

know may may have may get begged and you

know we may see both court casees as

well as you know opportunities to

regulate differently to say u you know

why shouldn't someone be able to um

advertise

um as a clinician on a site like that um

when when you know that's just sort of

not the world we live we live in

Um because you so then you're left with

what like pharma devices, supplements,

you know, u plenty of things. Um uh but

you know it is uh you know for

physicians maybe you know BMWs and

things like that you know that are that

are golf memberships. Now now you can

edit all that stuff out

but um

I I I like I admire the fact that they

didn't address that now. Um, I I think I

think most people expect information in

healthcare to be free. [clears throat]

Whether or not it just feels different.

It feels different than other types of

information. And I can't give you a real

business explanation of why it should be

free. Um, but it feels like it should be

free and I think it was a good decision

uh on their part. And [clears throat] so

what do you do with that is is your

question. And I think um there probably

are lots of smart people around there

figuring out whether there's enterprise

models and other other models and so

forth that are going to you know carry

some of the some of the water. But um I

kind of you know there there's a pretty

good track record um you think of Red

Hat and think of other types of of of

things where people have said you know

we don't have to charge for the thing.

um we can find ways to to to to build

around it and I and I I'm not worried

about that and so neither is a lot of

smart money.

Well, Andy, thank you so much for coming

on. This this was amazing and we'll see

we'll see what happens next year. Uh

it's it is it is the most exciting time

for healthcare right now and really

appreciate you coming on. Well, you guys

are really on the cusp of it all. And I

so cool to be on this show. Um, where

you guys are where I think, as you said,

Justin, there's just so much interest

and so much focus. Uh, and everybody's

dying to know uh what you guys are are

able to show people. So, thank you for

doing all that.

Yeah, thanks for coming on, Andy. Really

appreciate having you on. We are the

final guests of the year. So, we're

we're we were maybe going to wear a

Santa hat this time, but we didn't we

didn't do it. Uh for the end of the

year, maybe a New Year's Eve celebration

of some kind. But, uh thanks for joining

us. Happy holidays.