Latest / AI for Founders with Ryan Estes / AI Can Predict Disease Years Early

Transcript

AI Can Predict Disease Years Early · AI for Founders with Ryan Estes

0:00
Welcome to AI for Founders. I'm Ryan Estes.
0:02
What happens when you build an AI that can spot disease years
0:05
before symptoms? That's the problem I guess today
0:08
ran straight into on this episode of AI for Founders.
0:11
I'm talking with Mariano, four time healthcare entrepreneur and
0:15
CEO of Xenia. Xenia is an AI powered operating
0:19
system for corporate health that monitors over 100,000 people,
0:23
runs around 95,000,000 clinical inferences a month, and can
0:27
intelligently route people to care across more within 40
0:30
conditions. If you're a founder building an
0:32
AI health tech or any regulated incentive broken industry, this
0:37
episode is for you. You're going to learn how
0:39
Mariano redesigned his entire business model so someone
0:43
actually had financial reason to pay for prevention, and how he
0:46
thinks about AI as a copilot for doctors instead of replacement.
0:50
And how he keeps his sanity and creativity through writing,
0:54
photography, sports, and reading while running his fourth
0:57
company. If you want conversations like
0:59
this distilled into your Founder Ready experiments every week,
1:02
head to AI for founders.co and subscribe to the newsletter.
1:06
If you want the real stories and playbooks from people actually
1:09
shipping AI products, not just tweeting about them, go to
1:12
aiforfounders.co and subscribe to the newsletter.
1:15
And is it tweeting anymore or is it Xing?
1:17
What are you doing over there anyway?
1:19
And if you want to feel a little less alone building weird
1:22
ambitious things in public, you guessed it, newsletter at AI for
1:25
founders.co. And if this episode helps you
1:28
pay the favour by tapping 5 stars or dropping a short review
1:31
in your podcast app so more founders can find the show.
1:34
Now a quick word from our sponsor, Kick Caster.
1:37
If you're a founder, you already know the hardest part is not
1:40
having something valuable to say.
1:42
It is getting the right people to hear you say it.
1:44
That's where Kick Caster comes in.
1:46
Kick Caster is the agency I Co founded that books founders,
1:50
experts and executives on podcast their buyers already
1:53
listened to. Here's how it works.
1:55
My team builds a custom interview road map around your
1:58
story. Then we line up interviews on
2:00
targeted shows in your niche, handle all the back and forth,
2:04
and make sure you show up with angles that drive leads, hires,
2:07
partnerships, or the next round. Our clients have landed on
2:10
hundreds of top podcasts, generated millions in revenue,
2:14
and turned a single interview into content for months.
2:17
You do what you do best, show up and talk about the thing you're
2:20
obsessed with. We do the unsexy work of
2:23
research, outreach, and scheduling so you can build
2:26
authority at scale. If you're ready to turn
2:29
conversations into pipeline, go to kickcaster.com, apply in a
2:32
couple of minutes and tell them you came from AI.
2:34
For founders that's Kick Caster by MO Burst.
2:55
All right, Mariano, welcome to the show.
2:57
Thank you so much for joining me for this conversation.
3:01
Thank you so much for inviting me.
3:03
Absolutely. My pleasure.
3:04
I'm really looking forward to speaking with you because you
3:07
seem to be a man on a mission and a man of principle.
3:11
Not to mention you. You have beautiful photography
3:13
and hopefully we can get to that as well, but why don't you begin
3:18
by giving us an introduction to your company and the the
3:22
customers you're serving with with that company?
3:24
Sure, sure. Axenia was was born out of sheer
3:29
curiosity, mostly was five years ago.
3:32
I was, I came from a series of other healthcare ventures and I
3:39
wanted to explore a little bit on where, you know, the
3:42
increasing amount of data that we were having and, and AI,
3:47
which I think are the two things that are, are, are, are changing
3:50
or revolutionizing healthcare at this, these days.
3:52
What, how that could be used? This is, remember, this is five
3:55
years ago. So AI is not what it is today.
3:59
And we, we had a little data, but not as much.
4:01
And our logic was healthcare has been changing their targets for,
4:08
for a few decades now. It, it, it was born in, in its
4:12
modern form around 1940. And, and at that time, most of
4:17
the issues were had to do with, with acute diseases, with
4:20
infections and, and, and things like that.
4:23
And we used, think about it, you know, over the second-half of
4:27
that over the last century, we were, we switch our focus to
4:31
chronic diseases, which are quite different than, than acute
4:33
diseases. And the way they're, they have
4:35
to be treated, for one, they, they have almost almost no
4:39
symptoms or very little symptoms up until they say it is very
4:42
late. And remember for, for, for, for
4:45
the prior generation of diseases, symptoms are the one
4:49
thing that takes you to doctor. So when you're, you're, you're
4:52
banking on that to take people to the doctor and you don't have
4:54
symptoms, you have an issue there.
4:56
And the other, the other thing is that, you know, treating
5:00
chronic diseases just takes a lot of time and, and you have to
5:04
do many complex things and without much of A, of a feedback
5:10
system. So we were excited that, that
5:12
you can do two things with data. You can, you can collect it very
5:15
continually. So, so you can understand, you
5:19
know, how your blood pressure works not only at the doctor's
5:22
office, but also when you're watching TV or playing with your
5:25
kids or different activities during the day.
5:27
And there's no reason why we cannot measure that in a more,
5:30
in a more continuous matter. And then the other thing is we
5:34
can, we can examine data in, in much wider quantities than
5:39
before. So you, you can, you can, you
5:43
know, get dozens and, and you know of, of data points, just
5:48
put it on the web and analyse it every minute for every patient.
5:51
So you have a chance of actually to bringing patients to the
5:55
doctor earlier. And that means that you're going
5:57
to spend less and they're going to be better treated.
6:00
And that's, that's where we started.
6:01
And so basically we found out that it was possible to monitor
6:06
patients and intervene earlier. We actually took about took us
6:10
about a year to do a big clinical study on that.
6:14
And we proved that we could lower the effects of of diabetes
6:19
in your heart and in your brain and in I mean the overall death
6:23
by 20% or something like that. And at the same time lower,
6:27
lower expenses. So that's, that was kind of our,
6:31
our, our, our, our start. And from there, we, we put
6:36
together a much larger models that now now can identify
6:41
something like 40 something different diseases and take you
6:43
to doctor before and help you understand where you have to go
6:48
instead of bumping around in different doctors and things
6:50
like that with, with, with intelligent navigation.
6:53
And we are, we are actually getting good results.
6:56
And the, you know, in a, in a, in a market that normally has a
7:02
rate of increase of around 2.5 times the inflation, we are
7:05
getting something close to inflation is still a little bit
7:08
higher than inflation. Our, our target is to, to take
7:10
it lower, lower down to inflation.
7:12
But it's, it's a big, it's a big saving.
7:14
And we feel that it's going to have, you know, huge impact
7:18
when, when this would actually grow.
7:19
So it's already having some impact in in Brazil, but you
7:22
know, this is a global problem. Incredible.
7:26
So Xenia is gathering data points from from patients or
7:31
from from absolutely anybody. What?
7:33
What data is it actually collecting?
7:36
It's mostly three types. We, we get the financial data,
7:40
which is the easiest to, to collect because we are plugged
7:43
into the, into the insurance system.
7:45
So we know when, when you go to a doctor, we know you've been
7:48
there basically because you've paid for the doctor.
7:50
So we, we track the payment or you, we know that you bought
7:53
something in, in the pharmacy or you did a little, some exam or,
7:56
or, or whatever things they do that way.
7:59
So that's, that's kind of our, our, our, our backbone because
8:02
you that there you can have personal information.
8:05
Then we have clinical information that we captured
8:07
through different methods. Wearables is one thing and but
8:12
you know, wearables in Brazil you have a penetration of around
8:14
15% of total talk relation. So then we have we have other,
8:20
other methods. We have a face scan that
8:23
actually takes a picture with your own cell, cell phone.
8:27
You send us a picture, actually a video, 32nd video.
8:29
And we can infer from there some 32 different clinical variables.
8:35
And and then we use, you know, basic, basic questionnaires as,
8:40
as they as of now they have been using medicine for years.
8:43
So, so we, we gather other other other than the data types from
8:47
there. Some patients share with us the
8:51
results of their labs and then we, we, we capture that as well.
8:55
We capture data from labour health, which in Brazil is
8:58
obligatory, is mandatory once a year.
9:00
So we capture data from there as well and we capture there from
9:03
partners. So when, whenever we, we put
9:06
together a system to to cover our population, you always have
9:10
like the, the big piece is a, is a health plan, but then you also
9:14
have small other partners that plug there, you know about an
9:17
obesity program, for example. And then we capture data from
9:20
them as well. And, and we have our own, our
9:23
own navigation nurses that cover something close to the five, 5%
9:29
of total population, which is the most, the most critical 5%.
9:33
And then for those people, we capture a little bit more and
9:36
everything goes together into a beta lake.
9:39
Amazing you talking about so, so the financial information you
9:43
can keep track of people are on target with their their renewal
9:47
of their prescriptions that they're seeing their doctors,
9:50
you know, you have clinical information.
9:52
So you can kind of track like what diagnosis are what
9:54
treatments are kind of keep track of that.
9:57
And then there's a wearable component.
9:59
When you mean wearable, you talking like an Apple Watch or
10:01
something like that. If you have an Apple Watch, we
10:03
would we would pluck to that. If you have any other other
10:06
wearables of that sort, we'll pluck to those as well.
10:11
Glucose monitors, things like that.
10:13
Glucose monitors, it could get your heart rate, it could give
10:17
blood oxygen content. You have all those data points.
10:20
If you, if you have a wearable, we have a lot of data from from
10:23
you, but, but unfortunately you know, there's still things
10:26
particularly in a country like Brazil where there are so
10:28
somewhat expensive. So it's it's getting better, but
10:31
the penetration is is still small.
10:34
They're expensive in the States too.
10:36
It ain't. Cheap, they're going down in
10:39
prices, they're going down in prices.
10:40
The most, the most famous one are still in the hundreds of
10:44
dollars, but but you you have some some in the lower 40-50
10:49
dollars. This episode is brought to you
10:54
by MO Burst, the AI first growth engine trusted by iconic
10:58
startups like Calm, Robin Hood, Reddit, Uber and Google.
11:03
Moburst is a digital, full service, mobile first marketing
11:06
agency that helps companies to scale and become category
11:09
leaders. Whether you're finding product
11:11
market fit or ready to accelerate growth, Moburst
11:15
delivers creative, data-driven solutions that help with scale
11:18
and efficiency and impact. At Moburst, we're not just
11:22
guessing what works. We're predicting it.
11:24
As an AI first agency, Moburst uses cutting edge tools to
11:28
analyze user behavior, adapt creative in real time, and cover
11:33
hidden growth opportunities. The result?
11:35
Marketing that's faster, smarter, and built to convert
11:39
from early stage startups to global brands.
11:41
We've helped hundreds of companies overcome complex
11:44
challenges, unlock new audiences, and optimize every
11:48
metric that matters. At Moeburst, we believe that
11:51
creative thinking, advanced technology, and data are the
11:54
core drivers of success at scale.
11:57
If you're serious about scaling your business with precision and
11:59
purpose, it's time to talk to Moeburst.
12:02
Visit moeburst.com and let's build something category
12:05
defining together. OK, I've got a confession.
12:10
As a podcaster and founder, I'm constantly meeting incredible
12:13
people, guests, listeners, or other entrepreneurs.
12:16
But here's my little secret. I'm terrible at following up.
12:20
My inbox is basically a graveyard of missed
12:22
opportunities. Does that sound like you?
12:24
You meet someone amazing, exchange emails, have great
12:27
intentions, and then life happens.
12:29
Six months later, you're kicking yourself thinking I should have
12:31
stayed in touch with that person.
12:33
This is exactly why I started using Warm Star dot AI and
12:37
honestly, it's transformed how I think about my network.
12:40
Instead of letting relationships go cold, Warm Start analyzes my
12:44
existing contacts and helps me identify who I should be
12:47
reaching out to and why. It even writes the outreach
12:50
messages for me. But here's the key.
12:52
I review everything before it goes out.
12:55
I'm not just talking about this because they're a sponsor.
12:57
I'm talking about it because watching a healthcare executive
13:00
use this to build his entire Advisory Board in one weekend
13:03
before CEO sign ups made me realize I was leaving money on
13:06
the table every single day. Your network is your net worth,
13:11
but only if you actually nurture it.
13:13
Check out warm start. dot AI. Use code AI IFF to get 50 bucks
13:18
off your first month stop letting your relationships turn
13:21
into miss revenue yeah. You know, this, this is
13:24
something that I'm very, very curious about in regards to
13:27
Healthcare is all these different biometrics, all these
13:30
different ways we can assess symptoms, you know, and I think
13:35
where AI is particularly adept because it, it can, it can
13:40
process so much more data in parallel.
13:43
And what I'm curious about is because when I was a teenager, I
13:47
worked at Whole Foods Market, which is like a, you know, a
13:50
grocery store, whatever. I was a produce guy.
13:52
So I was a guy stacking up the carrots and apples.
13:55
But in the, the natural living department is where they sold
13:58
the vitamins and things like that.
14:00
And there was a gown named Pammy.
14:03
And Pammy was a natural path doctor.
14:06
She was a nutritionist and kind of a witch in the best possible
14:11
way. And she was working basically
14:13
semi retirement at Whole Foods, you know, selling vitamin C and
14:17
echinacea, you know, but people would come in, you know,
14:20
particularly people that are of like a, a natural bend, like
14:24
they're not going to go see the doctor.
14:25
So they'll like literally bring their cancer diagnosis into this
14:29
case, Pammy. And she had this unbelievable
14:32
knack of being able to treat the symptoms naturally for them.
14:37
But what she used to say is like, I can smell it on them.
14:40
I can smell the different disease or things like that.
14:44
And it, it always struck me largely because she had like a
14:47
devoted following of people that would come to her for medical
14:50
advice. But it seems like we're getting
14:52
closer and closer to that with these devices.
14:56
So, you know, right now I don't know if like wearables or phones
15:00
can smell anything, but they can take a picture.
15:03
I'm really curious about when somebody's using Xenia and they
15:07
take a picture of their face. What is it measuring?
15:09
What kind of feedback is so So what we're using?
15:12
Is a model that was was developed by MIT.
15:15
It's actually that we, what we do is that when you know, this
15:18
is published science data, they, we, we are measuring the colour
15:23
of the blood vessels in your, in your face.
15:25
And that gives us an idea of glucose, give us an idea of, of
15:32
cholesterol and, and, and a bunch of other stuff, blood
15:36
pressure as well and things like that.
15:37
None of that. It's a it's, it's supposed to
15:42
replace a proper diagnosis. This is just, it's, it's just a
15:46
guess the way, the way we, we, we look at it, it's, it's, it's,
15:49
it's a good guess that, you know, this is a person that
15:52
should go to the doctor, then check it's, it's blood pressure
15:54
or, or, or, or whatever. So, and what we have we, we, we
15:58
use kind of layers and, and I think your, your point is, is,
16:02
is, is, is the one thing that that's important.
16:04
And it's so difficult to process the amount of data that we have
16:07
today that unless we took, you know, approaches like these, we
16:10
wouldn't be able to, to, to, to cover the whole population.
16:14
And so, but, but you have to be careful the way we're not making
16:17
an, you know, a diagnosis that the person has high blood
16:20
pressure. What we're saying is the chances
16:22
that this person having a higher pressure is average or more than
16:26
average. And so we we actually send those
16:29
first to one of our nurses and the nurses probably would take,
16:31
we will send them to a doctor. Credible I I would be afraid
16:36
that if I took a picture with Xenia, just immediately call
16:39
like an ambulance. No, we don't.
16:42
We it doesn't work away. No, it doesn't work away the way
16:45
the way. I mean, it would be fantastic if
16:48
it could. I mean, some dogs, some, some,
16:49
some, some somewhere I watch it would do that and the whoop
16:53
would do that. If they, if they detect an FDA
16:55
feed, they would actually call an embryos for you.
16:57
It's, it's not what do we, what we do is, is, is a little bit
17:00
more long term. So we are, we are measuring the
17:02
trends and the chances of, of you having trouble in the
17:06
future. And that's, that's the most
17:07
important part because we want to capture you, not when you're
17:11
having a heart attack, but we know hopefully years before, so
17:14
that we can do things to prevent that from happening. 100%, you
17:18
know, and the, the data points will keep emerging.
17:21
There's a company called Canary Speech and I spoke with the
17:24
founders friend of the show and they're doing something similar,
17:27
only it's all with voice. So what it does is it keeps, it
17:31
listens to you and it can detect whether you have Alzheimer's or
17:35
you have dementia. You, it can, it basically is
17:39
reading you from all these other data points.
17:41
So you know, we have, you know, we have voice input, we have
17:45
photograph input. You know, as you're looking into
17:48
the future, you know what kind of but biometrics and what kind
17:52
of data would be kind of the Holy Grail like the for
17:55
diagnosis kind of from where we're standing right now.
17:59
It's difficult to say. I mean they they're attacking
18:00
the problem from from different angles.
18:02
I'm biased. I believe that the Holy Grail is
18:06
blood that we got. We are we're getting our chances
18:09
of developing biomarkers in the blood and there are many
18:12
companies that are doing that in the US already.
18:16
Prices are still very high to to deploy those on a on a
18:19
population basis, but they're coming down.
18:22
But I believe that that you know, with, with a few drops of
18:25
blood, you can, you can actually get a lot of biometric data and
18:30
from there from there very, very accurate predictions on, on
18:34
what's going to happen. I think that we are seeing those
18:37
things. I mean, I think that warp
18:39
actually is, is, is tying that up to their to the wearable.
18:43
So you're going to have both the wearable that will end the
18:46
biometrics to, to take care of that.
18:48
So I, I think that's, that's going to be the future.
18:50
Now WHOOP is a very high end, very expensive solution.
18:54
And we need things that we know we can deploy to the whole
18:57
population, but not not only for the rich, but I think, you know,
19:01
prices are prices are coming down.
19:04
Yeah. Is there a little bit of cold
19:06
feet as far as like single drop blood analysis after the whole
19:10
Elizabeth Holmes debacle? I think if you if you, if you
19:15
wanted to do exactly what she was doing, I think they would
19:18
be. This is a different thing
19:19
though. What what they're capturing is
19:22
and, and, and I didn't mean really naturally 11 blood drop
19:25
of blood if it's a small amount of blood, but it's not one drop.
19:30
However, there are companies that are trying it.
19:32
They're trying to to succeed where she where she failed.
19:34
Actually saw a couple of companies to it trying to get
19:37
the data from a from a few drop of blood.
19:39
Literally, yeah. I mean, I have faith they'll
19:41
figure it out. You know, I mean, it's coming
19:43
for sure. Now, as far as like, you know,
19:46
people that have acute pathology and, you know, disease, this is
19:49
a great product. What about folks that are feel
19:52
good, want to feel better, you know, well, treatment for the
19:55
the the healthy. Do you see like an Ave. there?
19:58
There's clearly it's, it's not what we're focused.
20:01
We are, we are more focused on, on, on, on anticipating.
20:04
Large problems in the population that would cost the significant
20:07
money. I think there is definitely a a
20:09
huge amount of of of people worried about living longer and
20:14
living better those years as in the market is, is is very
20:17
interesting. I think that, you know,
20:20
unfortunate as though all new markets, I mean there's a
20:22
there's a mix of things there. So there are there are more,
20:27
more more proven things and less proven things at this point.
20:31
So it's a little bit of a of a mix at this point.
20:34
But I think there is there is some interesting, very
20:36
interesting market there. We all we all you know hook to
20:39
it in some way or one way or the other.
20:40
I think it's. Yeah, it's, it's definitely the
20:44
future when you're seeing like a, an evolution of the product
20:47
where it becomes, let's say like 100 X better at diagnosing
20:52
potential problems. How do you see like the, the
20:55
role of your doctor kind of evolving if it's if it's not
20:59
necessarily doing all the the prescriptive needs?
21:01
I mean, we're not there yet, but where do you think the role of
21:05
the doctor will be in the future?
21:07
I I don't see. In the, in the, I don't know,
21:10
it's very difficult to to project 20 years out, right.
21:12
But I don't think the next 10 years that doctors are going
21:15
anywhere. I believe that they're getting
21:17
much better in choosing the patients.
21:19
I think that there, there is a big inefficiency in the way we
21:22
choose patients today. And by I mean, I'm pretty sure
21:24
that you and I have gone to the doctor many times and you, you
21:28
know, you went, you wasted an hour of his time and you left
21:31
there very happy because you had nothing and, and you are, you're
21:34
perfectly healthy. And so that's an hour that that
21:37
he lost, you know, looking for a patient that actually needs it.
21:40
So finding the right person for the doctor is, is one thing that
21:43
I'd actually see that that we're getting better and helping the
21:46
doctor reach diagnosis quicker. It's also it's also true.
21:50
So we have, for example, a copilot for our nurses that
21:53
actually allows them to treat patients.
21:56
But I don't think that that we will ever be able to replace
22:00
them. I think people will not have the
22:02
same engagement. And you know, this could this
22:04
could change, right? So many generations from now
22:07
know what, what do we know what, what age size can do?
22:10
But at this point, I don't see the patients actually feel
22:14
confident enough to be treated just by an AI.
22:17
So the combination of an AI and a human is very, very powerful.
22:20
But if you can save the time of that human for the real cases
22:24
and you save the time, you know that they spend asking questions
22:28
or, or or or or coming to to a diagnosis that that can be done
22:33
much easier for a machine, I think that's that's a big
22:36
improvement. Yeah, agreed.
22:38
I mean, you want to go get some blood work done, it takes 1/2
22:41
day, you know. Yeah, at least in the States,
22:44
those those gals get that are drawing blood.
22:46
Not that friendly, you know, but that way, well, you can.
22:50
You can probably do that at home now, right?
22:52
I think here, at least in Brazil, is is very common and,
22:55
and it's but it's cheaper. Fun.
22:56
Funnily, I don't know, I'll build it in a temp data.
23:00
I'll have to look into it because if it'll save me 1/2
23:03
day, you know, I'm definitely in, yeah.
23:05
Yeah, it's, you can probably do it in like 10 or 15 minutes.
23:08
And they're, they're very, very, I think frankly, it's cheaper
23:11
than actually going to the to the hospital.
23:14
Yeah, there, there seems to be a little bit of conflict as far as
23:18
AI and medicine. You know, the, the medical
23:21
establishment in the States moves notoriously slow and maybe
23:25
for good reason. But I've kind of heard that like
23:28
as we're introducing, you know, these new diagnostic tools that
23:33
are there to aid doctors, we're, we're adding new data points
23:38
that doctors can actually assess that in fact, if they're getting
23:42
all these data points and let's say they miss the diagnosis,
23:45
that they could actually be held financially liable.
23:48
They could be sued if they missed the, the diagnosis based
23:52
on all of the information they have.
23:54
So now if we're overwhelming doctors with too much data,
23:58
they're almost like, hey, I, I don't want to look at all this
24:01
because if, if for some reason the AI does this or that, then I
24:05
could be held liable. Yeah.
24:07
Like how do we kind of overcome that?
24:09
Like that tension and and just like the the personal liability
24:13
from the doctors. I think it's an issue.
24:15
I think that we we have to have better models to to to do that.
24:18
I think that the the way the way we are approaching that as a
24:21
copilot where the doctor is, is the one taking the decision at
24:24
the end of the day. And we just giving them you
24:27
know, things that we saw it's it's it's probably safer way, at
24:32
least for now of doing that. I think that, you know, the it's
24:38
also very important that the systems are transparent.
24:41
So one of the things that we put a lot of emphasis in doing is,
24:45
you know, we think the model always tells us we think this
24:49
and this is the reason why we think that.
24:51
And these are the papers that actually support my reasoning.
24:55
And unless we can find that chain, we don't inform that to
24:58
the doctor. So if you design it well and you
25:02
put and you put a doctor in the end, I don't, I don't think
25:05
there is a reason for the doctors to be apprehensive of
25:07
that. But if you if you go directly to
25:10
the doctors, I don't think the models are that good yet.
25:12
I think there is, there is actually a study published two
25:15
or three weeks ago by, by a team of Harvard and, and Stanford
25:20
doctors and they found that, you know, the, the, the, the rate of
25:25
error in, in, in, in models by medical models, it's still
25:30
pretty high. Now we have to find a way of
25:33
coping that because the rate of, of, of, of actual medical
25:36
mistakes is also high. I think human, humans also make
25:39
mistakes. So we, we need to, you know, get
25:42
the best of both worlds. At one point.
25:44
We need to find a way of doing it.
25:45
But I don't think it's, I don't think it's, it's, it's, it's
25:48
irrational to to be be careful about it.
25:52
Yeah, 100%. Maybe talk a little bit about
25:56
who your customer is, who you're selling to and kind of where
25:59
your market is. Are you guys in the States now?
26:02
We're not in the States yet. We, we are in Brazil only.
26:05
It will, it will take maybe a couple of years for us to, to,
26:08
to go to the US. Still Brazil is pretty large and
26:11
that we want to make sure that, that we have a very robust
26:13
product before before we go there today.
26:16
What we do when this this changes country by country, but
26:19
the US and Brazil will work the same is, is we actually, we
26:24
actually take care of the insurance or, or we are ATPA
26:28
for, for the company. So what we do is we manage their
26:31
insurance system and we deploy the digital tools as pretty much
26:37
as we see fit. We don't charge for the digital
26:40
tools. What we charge is, is a, is a
26:42
fee if we actually lower their, their, their, their health care
26:47
cost overall. So that way, that way we feel
26:51
that it's easier for, for them to understand what we're doing.
26:54
And, and, and, and it's, it's, it's very important in
26:58
healthcare to be aligned with your clients in terms of your
27:01
revenue model. I think that one of the issues
27:03
that we see in healthcare in general is that there is a lot
27:06
of disconnects between the revenue model and, and what you,
27:09
what you actually have to deliver.
27:11
And so, for example, if you go to a hospital, I mean, you're
27:14
not saying that they do this on purpose, but the more you spend
27:16
there and the more you, they give you treatments for things
27:19
that are irrelevant, the more money they make.
27:22
Not I'm not saying that hospitals do that by any measure
27:24
of imagination, but they're, they're incentivized to do that.
27:28
And that's that's. That's say they do that.
27:31
Yeah, they're definitely. Maybe some hospitals even do
27:34
that. Yeah, on purpose.
27:36
Yeah. I mean, not to be like, I mean
27:38
pessimistic about it, but they definitely do, you know, And
27:41
it's it's. That's at least they don't avoid
27:44
it because it's it's in their favor.
27:46
Yeah, 100%. So you're getting paid on the
27:49
delta of saving with the technology.
27:51
Technology is free. So you have to you say, hey,
27:54
let's audit on the back end quarterly, monthly, annually and
27:58
see what what we saved and then you get paid a percentage of
28:02
that. Exactly, exactly.
28:04
And we do it. We do it on a yearly basis and
28:06
it's, it's fairly easy to met to to metric because what what we
28:09
measure is how much did you get of an increase versus the market
28:13
and those are very easy to measure numbers.
28:16
So there's it is very clear to to see how much we have.
28:19
Improved percent and I I believe in kind of researching for this,
28:25
this, this conversation, you guys have raised 12 million to
28:29
launch the the product. Well, we've raised more than we.
28:31
We've raised on Series 8 to 12 million in total.
28:34
I think we've raised something like 20.
28:36
About 20, OK, fantastic. And is that from state bound VCs
28:43
or where where kind of like your your partnerships coming from?
28:46
We, we have VCs from everywhere. We, we have Brazilian VCs and we
28:51
have American VCs as well. We have Chilean VCs.
28:54
We have an Argentinian partner as well.
28:57
It's, it's more of a global market, yeah.
29:00
You know it as I'm kind of getting ready for this
29:03
conversation, I really appreciate founders like you
29:06
because you're very outspoken and you have you're an author
29:10
and a photographer. As I mentioned, you've had, you
29:13
know, exits in the past. So First off, thank you for
29:17
that. I want to mention your book,
29:20
which is software is eating the world, yet healthcare remains
29:24
inedible. Talk to us a little bit about
29:26
the book and kind of the, the, the the reason you decided to to
29:29
pen it. I think the, the, the main
29:32
reason is that, you know, we, we have been talking about, about
29:36
transforming healthcare with, with technology for a long time.
29:40
I think that if, if you remember, I can't remember, the
29:42
year is more than a decade ago that Marc Andreessen wrote that
29:46
that famous self reset in the world memo for the for the Wall
29:49
Street Journal, It's maybe 12 years ago.
29:53
And if you go down 3 or 4 paragraphs in that memo, there
29:56
is there is a, a sentence that says, I believe that Healthcare
30:00
is Next up for fundamental transformation.
30:02
And you know, 10 years later it hasn't happened.
30:05
And parts partially there is, there is there is good reasons.
30:09
As you said, you know, the people are conservative in
30:11
healthcare. So you don't want to do things
30:13
that put people at risk. But I think partially is that
30:17
that issue of, you know, the huge amount of data and the
30:20
fragmentation of of the data and how you actually go through it.
30:24
And so I believe that we have come to the point that we can do
30:29
it. So, but because we have now we
30:31
have ability to, to capture that data and ability to to process
30:37
that data with AI. Just to give you a number there,
30:39
I mean we only have 100 and something 1000 patients in in
30:44
Brazil. And for those we do an average
30:48
of 95,000,000 clinical insurances each month to to be
30:54
able to do that, you know with with human based doctors, it
30:58
would take us a whole population of doctors in Brazil and we
31:00
probably won't be able to do it. So we now come to a point where
31:04
we have the tools to do it. So I wanted to, to, to show a
31:09
little bit of how can that be done?
31:11
What are the things that, that, that, that prevent us from now
31:15
from doing that? And what's our own experience in
31:18
changing that? Because it's, it's really, it's
31:20
really, you know, I don't want to say revolutionary for my own
31:23
company, but it's, but the impact is really big.
31:27
The, the our clients in in on average are spending are drawing
31:31
their expenses by half the rate of of the rest.
31:33
And you have to understand this is a, this is a market of about
31:37
10 trillionth globally and it's growing at 2.5 times the, the,
31:40
the, the speed of inflation. So it's going to become
31:43
unsustainable at one point. It's, it's, I never get closer
31:46
to that point. Unless we change the way we look
31:49
at healthcare, we, we're going to go broke.
31:52
Most countries are going to be not going to be able to pay for
31:54
it. So, yeah, you know, and it's,
31:58
it's really, I mean, the, the, the business model is just
32:01
fascinating and honestly kind of inspiring.
32:04
You know, I, I kind of don't like the fact that I am so
32:07
cynical about the healthcare system in in the States because
32:11
it seemed much more like a sick care system.
32:13
And you have to be skeptical. You have to kind of watch your
32:15
back with the insurance companies and seems like
32:17
everyone's out to get you, you know, and largely I think it's
32:20
kind of what you were talking about is that, you know, the the
32:24
value chain is misincentivized, you know, so the way you built
32:29
that business model to to basically be able to share in
32:34
the success, on one hand it's very inspirational, but don't
32:37
you feel like it's kind of a unbelievable risk as well?
32:40
Well, we're, we're very comfortable in the, in the
32:42
technology and at this point, 95 of our clients actually are,
32:48
are, are, are saving money. And that has been the truth for
32:51
the past three years, which we have been marketing this system.
32:55
So no, frankly, I believe that, that it is the way it is, the
33:00
way it should be done. I think that, that, you know,
33:04
companies have to put their money where their mouth is.
33:06
I think that it, it is, it is so common for, for every other, you
33:10
know, industry in the world. I mean, you, you don't expect
33:13
that you're going back to a, to a restaurant if they treat you
33:15
badly, right? So it's, it's the same thing
33:17
here. I mean, at the end of the day,
33:18
if, if, if they cannot keep you healthy, they shouldn't be
33:22
making money because they're not adding value.
33:26
I mean, I, I love that, but you know, you raise 20 million
33:29
bucks, you know, and I, I, I'll read a quote that I really liked
33:34
that, that you said saying have the, the definition of success
33:38
is having impact on other people building something that changes
33:41
live, even if it's small scale. Success isn't how much money you
33:45
make because that's a trap. Now how do you how do you sell
33:49
that to your your venture friends?
33:52
It's not a phrase I say a lot to the to the venture capital
33:55
community, to be completely honest, but.
33:57
That's between us, yeah. Yeah, between US and the 25,000
34:01
people that are looking, yeah, I, I, I actually believe that.
34:07
I think that, you know, I, I, this is my 4th company and I,
34:10
I've been lucky in my life. I, I, I wouldn't need to do this
34:14
for the money And, and frankly, I don't, I don't tend to hire
34:18
people that do this for the money.
34:19
I think that the company has to be sustainable.
34:22
But I think that, you know, there is, there is a lot to say
34:26
about having a purpose and, and want to want to changing the,
34:31
the world for, for the good. I, I, I believe about, I believe
34:35
in that, in that as a, as a way of improving our lives and, and
34:38
the lives of, of people around us.
34:41
But in any case, you know, I also, you know, even if you
34:44
didn't believe in that, even if you didn't want this to do this
34:48
for, for the good of it, I think that companies that sell
34:52
technology, they, they have to be held accountable for that
34:55
technology to work. I mean, you, you cannot, you
34:58
cannot get paid if you don't, if you don't have an impact.
35:00
And, and this is true in almost every industry out there except
35:04
for healthcare. Yeah.
35:07
I mean, you have to be reasonable on, on, on what
35:09
you're asking the people to do because we know that the
35:12
state-of-the-art is the state-of-the-art, right?
35:14
So even if we're the best doctor in the world, you know, there
35:17
will be people that get sick or get sicker after seeing them.
35:21
And, and we have to be reasonable about that.
35:23
We have to find good ways of measuring that.
35:25
But the current way, which is pretty much, you know, they have
35:29
no, no, no responsibility on the outcome.
35:33
I don't think it's a good either.
35:38
What are some of the differences between, let's say, the
35:41
Brazilian healthcare system and the American healthcare system?
35:44
The private system is very similar.
35:46
The Brazilians have a universal health system as well, Which it
35:53
covers everyone. And it's it's not horrible in
35:57
quality, but it's not the quality of what you would find
36:00
in a private system in the US. So you have that advantage.
36:03
Most people that that can pay for it would actually buy
36:07
private insurance. And that's about 30% of
36:09
population, still a large number given that Brazil is a large
36:12
country, but it's about 30% of total population.
36:16
And those people will have a system that it's compatible to
36:20
the US in terms of quality. And, you know, the kind of kind
36:24
of quality of doctors and things like that.
36:27
Yeah, the University of care is, is, is a little bit lower than
36:31
that. But, you know, it's, it's still
36:33
not bad. It's still not bad.
36:35
Nice. I, I do feel bad bashing them
36:39
all the time. Maybe I'm just spoiled because
36:42
I, I feel like we do have pretty good high quality of healthcare,
36:45
but boy, some of the bureaucracy is just really frustrating, you
36:48
know? But I think primary care is a
36:50
big issue. No, primary care is the issue in
36:52
the US. The access to primary care is is
36:54
is is is just too complex, right?
36:56
It may take you 2 weeks to actually see a primary care
36:58
doctor. So I think, I think that's the
37:01
main issue when you when you specialty care is pretty good in
37:03
the USI think it's primary care that actually lacks.
37:07
Yeah, but you know, a frustration tomb while I'm
37:09
airing grievances here is some of the more some of the more
37:16
preventative care or just like kind of like health and
37:19
prosperity kind of stuff, whether it's nutrition or diet
37:23
or training or some of the other therapeutics.
37:26
If you're talking about what's, you know, kind of tip of the
37:28
spear stuff, peptides, things like that doing, you know,
37:34
massage therapy and chiropractic and acupuncture.
37:37
I, I really kind of align with that kind of thing of like
37:40
investing in Wellness as opposed to just straight healthcare.
37:44
And it still seems like pretty divided, you know, I don't know
37:47
if that's a product of like the education that they're getting
37:50
or what, but that's an area I'm hoping technology will help
37:53
bridge, which is to say that like, hey, maybe there's not
37:56
clinical evidence that carrot juice makes you see better or
37:59
something, but like we can see the benefits, but broadly
38:04
through kind of analysis of like Wellness care.
38:07
I think there is there are two things there, right?
38:09
So the, the the first one is how you structure the contract you
38:12
have with your, with your insurance.
38:15
So if your insurance contract is very, is very short, which is
38:19
the case mostly in, in, in everywhere is, is you normally
38:22
have an insurance that you buy for a year and then you have to
38:25
renew it. And what happens there is
38:29
biological processes are much longer than that.
38:31
So if you, if you have an insurance company that says, you
38:35
know, I'm only having this guy for about a year, will I want to
38:39
invest in things that are preventative?
38:41
And the answer is, in most cases, it's no because, you
38:45
know, I know that if I help you, you know, whatever, lose weight
38:49
or exercise more or things like that, everyone knows that you're
38:52
going to be more healthy in the future.
38:54
And everyone knows that you will spend less money there.
38:56
There's no discussion there. The point is who is going to
38:59
spend that money in the future with you?
39:01
So if I spend money today to help you, you know, lose weight,
39:05
will I reap the benefits? And the answer in most cases is
39:08
no. So I think that's, that's,
39:10
that's part of it. The other part of it is, I think
39:13
that, you know, we, we've come to a point in the development of
39:17
technology where we probably have the people have to take a
39:22
little bit more, more care of themselves as well.
39:26
It's a little bit, I, I read it somewhere, I hope.
39:28
Well, I, I, I wish I could, I could remember where so I could
39:31
give them credit. But the, the logic, they said
39:33
just, you know, you don't expect your car insurance to pay for,
39:36
for gas or for, or for, you know, air for your, for your
39:43
bills or things like that. You pay for that yourself,
39:45
right? And then if, if, if you have a
39:48
major accident, then, then, then, then that's where the
39:50
insurance comes in. But in, in, in the health
39:53
insurance, we expect that the health insurance pays for
39:55
everything. There is no, no part of it that
39:58
we actually contribute a portion of that.
40:00
I think what we will see in the next coming years it's, it's you
40:04
know the patients will have to get a portion of the of the bill
40:08
as well as we do in every other insurance.
40:11
So I think those two things are, are are the ones that actually
40:14
prevent them from from seeing this things with a with a better
40:17
light. And then then the whole thing,
40:19
if you know you don't have perfect data for this or
40:22
whatever it, it, it becomes more of an excuse if you, if you
40:26
actually don't, don't have the, the incentive to do it right.
40:30
Yeah, I want insurance provider. They'll give me robot insurance,
40:34
which is to say when my hands give me a hard time, they just
40:37
hack it off and give me like a cool robot.
40:39
Hand Yeah, yeah, they would. I would love as well that'd.
40:43
Be great, Mariano. You're a very creative person.
40:46
You wrote a book, your photography's gorgeous, you're
40:50
out here doing interviews. Why is creativity important to
40:53
you on your your 4th company? Yeah.
40:55
I don't, I don't, I don't know. I, I, I, I think I, I never see,
41:01
see this as connected, but someone asked me the question
41:05
last week or so. And there are a few things that
41:08
you just do because your brain needs it, right?
41:11
So I'm not sure I would be able to be a, a, a decent
41:15
entrepreneur if I didn't do those things.
41:17
Are there other things that, you know, clear my brain for
41:20
thinking broadly? And so I, I cannot put the
41:23
finger on what exactly that helps me, but I think it
41:27
creates, it creates space in your mind in a way.
41:30
So I, I think I, I do those things more for that, more for
41:33
the space in my mind than than than than because there is a
41:36
link, the direct link of creativity and and
41:39
entrepreneurship. Yeah, we have such a hustle
41:43
culture. I don't think that gets
41:44
mentioned enough. What else do you like to do to
41:46
kind of create that freedom of mind?
41:48
Oh, but it's, it's mostly sports and I read a lot and I, I write
41:52
and, and in photography. Yeah, that's, that's that's
41:55
that's pretty much it. Thanks.
41:58
All right, it's that time in the podcast where we're going to hit
42:00
the lightning round. Are you ready for this?
42:02
Yeah, absolutely. OK.
42:04
You're obviously very bullish about AI and technology.
42:07
Make the bear case. Where is AI not appropriate for
42:11
the healthcare system? Today I believe is not
42:14
appropriate to to treat patients directly.
42:17
I think that it's very appropriate to to to be a right
42:21
hand system for the doctor. I don't think it's appropriate
42:26
to treat patients directly for now.
42:30
If you were entering college today knowing what you know,
42:33
what would you study? I think the, well, I, I'm
42:39
suspicious to say, but I think the, the, the, the intersection
42:43
of technology and Healthcare is probably the most interesting
42:47
area in the market for the coming years.
42:49
So I would look at something like that.
42:52
Would you go computer science or would you go?
42:56
Yeah, knowing me, I would probably go computer science,
42:58
but that's that's a personal preference.
43:02
Nice. This audience loves ideas,
43:05
business ideas. If we're thinking about the the
43:08
kind of all the opportunity in healthcare and technology,
43:12
what's a business idea that you have that maybe you haven't
43:14
acted on yet? Well, I believe there there are
43:17
two areas of where, where there are inefficiencies in
43:20
healthcare. 1 is the one that that we are attacking.
43:22
The other one is the relationship between payers and
43:27
hospitals and, and places like that is very, very difficult
43:30
for, for payers to audit hospitals and to understand that
43:33
the, what they are being charged is, is, is the proper thing.
43:36
And, and, and to understand, you know, the differences between
43:40
the efficiency of one hospital and the other.
43:42
I think that, that AI has a lot to offer there.
43:45
And I think that's, that's a very interesting the area we
43:48
we're not, we're not looking at that, but I think that's a
43:50
that's a very interesting area. Would that be like kind of
43:53
competitive analysis between hospitals?
43:56
Things like that, yeah, fraud. Things like 25% of of of
44:00
healthcare expenses are fraud. Yeah.
44:03
So it's. Waste fraud in the detected.
44:06
Yeah, waste to brute abuse fraud.
44:08
Yeah, those things can be can easily be detected by by AI.
44:13
Totally nice. Is there ACEO or entrepreneur
44:17
that you closely follow or admire?
44:20
Well, everyone, everyone does Warren Buffett, right?
44:23
I think that that's definitely one.
44:26
He does how to drink a winner. Yeah, yeah, I think he's the
44:31
whole the whole way he's lived. He has lived his life up to now
44:35
and you know, the kind of lessons that he teaches on
44:38
purpose and, and what to do with money and things like that.
44:42
I think it's it's, it's very relevant.
44:45
On a personal level, he seems to be a very humble man.
44:48
He's very, you know, he's not not even conservative.
44:53
He's he's like frugal. Yeah, very frugal.
44:57
Very, very frugal. Yeah.
44:59
Amazing. What's something you know now
45:04
that you wish you knew earlier? I think that.
45:08
I think that what, what I, I think that the, the, the, the
45:13
thing that I learnt the most over my career is about, you
45:16
know, how to manage people, how to be more patient with them,
45:19
how to coach them into the team that you want to form.
45:22
I think that at the very beginning I was very much about,
45:25
you know, firing people and replacing them and doing things
45:29
like that. And that's very tough on, on
45:31
building a team. And I think that building a team
45:34
on the long run, it's very, very important.
45:37
And so I, I've, I've learnt to, to balance that a little bit.
45:42
There's kind of this sentiment of you hire slow, fire fast.
45:46
Are you saying that like to to kind of go against that
45:49
intuition? I would, I would.
45:51
Still, this would still hire slow I I think that I I don't I
45:55
don't fire people as fast as before.
45:57
Yeah, it has that worked out. Are you you?
46:00
Have you found that people can be redirected and?
46:03
They, well, it's interesting. I, I, I haven't found that
46:09
people can be redirected that easily, but I think that the
46:13
effect on the team, it's worth it.
46:16
So giving people chances off of learning and things like that.
46:20
Sometimes it doesn't work. But even if it doesn't work, the
46:22
effect on, on the on the team, on on the, on the whole overall
46:26
feeling of the team, it's important.
46:29
Yeah. And what you mean, what's
46:34
unsettling to you about AI? Well, it's, it's a whole, it's a
46:39
whole area. We don't, we don't know much,
46:41
right? So we all, we all have this,
46:44
this logic of I, I don't think that it's AI by itself, but it's
46:47
the, the, the way we are governing.
46:50
HI, the way we are, we are using data to train models and, and,
46:57
you know, just just taking data from, from, from people, you
47:02
know, without, without compensating them and then
47:04
things like that. I think that it's an area where
47:07
we're going to see a lot of wealth being concentrated.
47:10
And I believe that's probably not good in the long run.
47:15
And I think that that AI has to be governed in a, in a more pro
47:21
society way, But it's very early.
47:24
We'll we'll see how it evolves. Mariano, this has been
47:30
fantastic. I really appreciate your time.
47:33
Your company Xenia, which is a xenya.com.
47:37
People can check it out. Please keep us posted on how
47:40
you're starting to move into the states.
47:42
When you do move into the states, definitely hit me up.
47:45
I want to help with that launch. Really love hearing about the
47:48
product and what the future of healthcare looks like.
47:50
And I honestly, on a personal level, I really appreciate your
47:53
approach. You know, I, I feel like there's
47:55
a massive overhaul of the, the healthcare system is underway
48:01
and I, I love hearing about people that are in it for the
48:03
right reasons. You know, I, I want everybody to
48:05
be profitable, but I want, you know, the success of everybody
48:09
to be shared with the companies that are contributing to it.
48:11
So thanks for that. No.
48:12
Thank you. Thank you for inviting me.
48:13
It was great. A.
48:15
Pleasure. Where's a good place where
48:16
people can connect with you online?
48:19
LinkedIn has my profile, my my e-mail is there, but even my
48:22
cell phone I believe is there so it's very easy to get a hold of
48:25
me. Nice.
48:26
All right, well, I'm going to text you about getting some
48:28
tickets down to Brazil. Perfect.
48:32
Thanks so much for joining us. Thank you very much.
48:36
Hey founders, let me ask you this.
48:39
Are you building with AI or getting left behind by it?
48:43
If you're serious about scaling smarter, faster and bolder, you
48:47
need to check out AI for founders.co.
48:51
It's your go to resource for cutting through the noise and
48:54
actually using AI to grow your startup.
48:57
Each week, AI for Founders drops a sharp, no fluff newsletter,
49:01
dives deep with expert guests on the podcast, and hosts monthly
49:06
workshops where you can get your hands dirty with the real tools
49:09
and strategies. Whether you're precede or Series
49:13
B and beyond, launching something new or just staying
49:16
sharp, AI for Founders gives you the edge you need to win in 2025
49:21
and beyond. Hit AI for founders.co to
49:24
subscribe to the newsletter, tune into the podcast, and grab
49:28
your spot in the next workshop, and it's all free.
49:31
Build smarter, move faster, think bolder with AI for
49:35
Founders. Now back to the show.