July 30, 2025 • 47 mins
AI isn't just changing healthcare; it's providing the essential help needed to unlock a trillion-dollar opportunity for better care.
Andreas Cleve, CEO & Co-founder of Corti, steps in to shed light on AI's immense, yet often misunderstood, transformative potential in this high-stakes environment. Andreas refutes the narrative of healthcare being slow adopters, emphasizing its high bar for trustworthy technology and its constant embrace of new tools. He reveals how purpose-built AI models are already alleviating the "pajama time" burden of documentation for clinicians, enabling faster and more accurate assessments in various specializations. This quiet, impactful adoption is seeing companies grow "like weeds" beyond common expectations.
The conversation addresses how AI can tackle the looming global shortage of 10 million healthcare professionals by 2030, reallocating a trillion dollars worth of administrative work back into care. Andreas details Corti’s approach to building invisible, reliable AI through rigorous, compliance-first evaluation, ensuring accuracy and efficiency in real-time. He emphasizes that AI's true role is not replacement, but augmentation, empowering professionals to deliver more care, attract talent, and drive organizational growth.
Follow the hosts
Follow Atin
Follow Conor
Follow Vikram
Follow Yash
Follow Today's Guest(s)
LinkedIn: linkedin.com/in/andreascleve
X (formerly Twitter): andreascleve
Corti Website: corti.ai
Check out Galileo
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
I think the whole sort of animosity, like healthcare
doesn't like technology. Oh, doctors are are take
lackers. They don't get it.
I just think it's like super boring.
It's a little bit like telling NVIDIA 10 years in, like this
whole like GPU thing, like you've been pretty slow at it so
far. Is it really worth doing?
We are back on chain of I thought I am your host Connor
(00:22):
Bronson and today I'm delighted to have Andreas Cleeve, CEO and
Co founder of Corti with me. Andreas, welcome to the show.
Great to see you. Hey thanks for having me.
Today we're diving into a sectorwhere AI promises immense
transformation and where Corti are already replacing single use
apps with medical grade AI. That's right, we're talking
(00:42):
about healthcare. There's unique challenges to
actually using AI within healthcare, but there are
amazing opportunities. The field is often cited as a
high stakes environment where the margin for error is
virtually non existent. We're dealing with people's
lives. How do we responsibly and
effectively integrate AI into such a critical domain?
(01:07):
We often hear about healthcare as this scary place for AI, so
I'm excited to unpack everythingtoday with you, Andreas.
To start, can you give us a brief overview of where AI
adoption is currently here in June of 2025 in healthcare?
Yeah, it's a great place to start.
So healthcare, as most of laypeople, me included, knows of
(01:30):
it is, is riddled with language tasks.
So roughly 40% of salary dollar spend to spend on work that
relates to language. So obviously talking to
patients, the part we love all, especially us patients, but also
writing notes, going to HRS, writing letters, writing quotes,
invoicing. It is all the, the, the lifespan
of healthcare. There's a lot of language and
(01:51):
text tasks. And obviously we have a new
thing in the arsenal with all these the transformer based
models that are very, very decent at, at doing language
tasks. So what we're we're we're seeing
is a lot of adoption in how can we take some of the what like
the cold hands based language jobs.
That's like tapping away your keyboard when you should really
(02:12):
be be focusing on the patient and trying to see how we can get
these models to to do some of that work.
It's so interesting because I think my perception at least is
that there are barriers, regulatory or otherwise, to tech
adoption in healthcare. We may see surgical equipment
get adopted rapidly in some cases, but often there are
(02:36):
technologies that are coming outof Silicon Valley, for example,
that maybe take longer to actually get adopted.
Are we seeing that with AI, particularly with the capability
of general purpose models and the opportunity for them to, you
know, transform all these 40% oftasks you mentioned, how can
(02:57):
they reasonably meet clinical and compliance standards?
Yeah, it's, it's a, it's a fair question.
I'm probably going to rant now, but I, but I think I, I, having
lived in SF as well, I think this whole narrative about
healthcare being bad at adoptingtechnology, it's just really
boring. And it's a Riddle with lack of
ambition. And I mean, I get it if you have
(03:18):
you used to build like grade like SAS tools for your
marketing team, who as part of their work spending all their,
the computer and they're adopting stuff fast, coming to
healthcare feels super slow. And I hear so many people be
like, oh, that, that law sort ofruins this as an investable
class and opportunity. And it's like they won't adopt.
But if you, if you sort of stop for a moment and think about
healthcare, you go to your hospital, your clinic,
(03:39):
especially your hospital, but golook around.
How much technology do you see? I sit here at our offices in
Copenhagen where we have what, 70 researchers doing machine
learning and I would bet you that per person, there's more
devices in the in the healthcareinstitution in per per room, per
clinic than there is here. Totally.
So they adopt tons of technologyand they do it quite fast, but
they have a pretty high bar for doing it.
(04:00):
So I think the whole sort of animosity, like healthcare
doesn't like technology. Oh, doctors are or take lackers,
they don't get it. I just think it's like super
boring. It's a little bit like telling
NVIDIA 10 years in, like this whole like GPU thing, like
you've been pretty slow at it sofar.
Is it really worth doing? I think there's just such an
opportunity building something like life's worth worth that is
(04:22):
really compounding every year that is massive because
technology is just, it's a different game here.
It's a game about being taken for granted.
So like if you're a lying on your bed and you're a patient,
like, and somebody comes in and say, hey, we're going to do an
MRI scan. Like do you think it would be
cool? I'm sure you wouldn't.
So it's very facetious rhetorical question, but
obviously you wouldn't be like, Hey, I want to see your ISO
(04:42):
standards first. So in a universe where you
actually want to take for granted that everybody is up to
their absolute best, they're delivering the best work they
can do. There's the lives work.
Every time they meet you, they are doing the best thing with
the best tools that are most applied to to your case and
they've been tested into oblivion.
If that all has to be true, thenwe can't also, while we're
(05:04):
sorting like kombucha mode and we don't feel any pain, be all
worried about oh, like healthcare, will it ever change?
Healthcare doesn't necessarily need to change.
It needs help. And that's, I think sort of one
of sort of the hidden secrets here is like Healthcare is only
for a, for a, for a boring, stubborn people who actually
want to have impact and want to be taken for granted.
I love the passion you bring to this because you're completely
(05:28):
right, there is a, a lack of ambition as you put it, where
often we have another B to B sass product then oh, I can't
get into healthcare because it'snot needed there or it's, it's
not a great fit or simply because they don't want to deal
with personal identifying information and healthcare
standards around that. And yet there is so much
(05:51):
technology in healthcare today. And I know you have some great
statistics about where the rubber meets the road here.
How many physicians, or anyone for that matter, are using AI on
a daily basis in the medical field?
So what are the real world outcomes when using purpose
built models like Cordy? Yeah.
(06:13):
So, so when the first generationof AI models build on us are
general purpose stuff, we're seeing a lot of pick up in
certain parts of healthcare. Now we get a little bit tedious
and boring here, but a lot of people think about healthcare
this big sort of monolith. It's definitely not.
And then they look at some of these startups and some of them
are amazing companies like, well, one of my favorites is a
Bridge American fantastic company who offers AI scribe and
(06:36):
AI assistant and copilot technology.
They've built an entire platformto to do some of the best
physician augmentation. They're seeing like any SAS
founder would be envious of their their kind of growth.
And they're like really, really good at outpatient care.
There's also really cool companies really good like rad
AI American company doing a documentation augmentation in
(06:56):
radiology. They're a super cool company.
They're really deep in inpatientradiology.
I guess some outpatient as well,but they're focused elsewhere.
We have a lot of like we just saw a fantastic IPO of Hinge.
I'm sure Hinge is using tons of AI.
There's a completely different workflow.
So the point is just the rubber meets the road a little bit
differently everywhere because Healthcare is not just one
monolith trillion dollar admin market, it's 1000 billion dollar
(07:18):
opportunities. So it's a very, very different.
But what we do see a lot is thatin cases where there is a very
large language load to get mandate, like imagine you're my
doctor and you want to make sureyou're sending me somewhere
else. That might be to our pharmacy or
it might be some more treatment to make sure you're doing the
right thing. You have a host of things you
need to take care of insurance. Am I covered or not?
(07:42):
What's the right path? What's the right medical data,
clinical data? Is there any research I need to,
to just double click on and what's the procedures that
usually offered to somebody likeme with the picture of symptoms
I I'm presenting? All of these things has to go
into sort of 1 assessment. And in these cases, usually we
see a lot of notes and a lot of note bloat.
(08:03):
In these cases, AI is really meeting a hungry group of users
who's spending way too much timelate at night.
It was like many called their pyjama time thinking about what
happened at 2:00 PM today and what did I say and what
happened. I had leave me meet a lot of of
healthcare professionals who arein those cases where they did,
they had back-to-back to back all day and I was sitting at
night. Actually, they should be
spending time reading out stories to the kids, but they're
(08:24):
sitting there tapping away at the keyboard.
And those are the kind of cases that's really working, that's
getting picked up where you can get a far, far along away with
with the general purpose models out there.
But we're just scratching the surface though.
Yeah, I love this description ofhealthcare as 1000 billion
dollar opportunities because you're right, so many outsiders.
(08:47):
Do you see healthcare as a monolith despite there being a
ton of differentiation? How would you advise
technologists who are interestedin the space, who maybe are
looking to drive AI adoption, torethink their perception of
healthcare as an industry to address these diverse markets
(09:08):
that you know are more granular across the the entire endoscope
of healthcare? I think the, the first thing
comes down to obviously wanting to do it.
It's not easy work, but that's the entire point.
It's not supposed to be, it's supposed to be really hard, but
it's really rewarding to do whenit's working.
And obviously there's some fantastic companies building
healthcare and health tech and health and being among them is a
(09:31):
very profitable opportunity. So there is both the, the, the,
the, the personal part and there's obviously the, the, the
vision impact part. So First off, we need to be
willing to spend the time. Secondly, I think it's important
to remember that when we sit here and we enjoy spending time
in design and we discuss what new cool design tools can I
augment my feet mouth stuff with.
(09:51):
It's a core part of our designers workflow to think
about it, which tools they're using.
It's not a core part of work fora physician.
Hopefully to think a lot about which part of the plug insurance
from our EHR providers have my bosses, bosses bought, bought.
They should be thinking a ton about that.
Obviously, hopefully they get what they need.
The point is the, the true goal here is not for you to to be
(10:12):
sort of on the dance floor. It's to make sure the people who
are on the dance floor, they have the most great time on the
dance floor. And that's very different from
other SAS products, which, whichare much more sort of high
identity product for the users. So I interviewed tons of
salespeople and many of them arelike very religious about do I
use Salesforce and hotspot? Obviously that happens in
healthcare too, because switching costs are high.
But it's not because I like the physicians we meet take like a
(10:35):
great pride in being an epic person versus being a certain
person. It's because they're good at it.
And I think that that's sort of a second part that that's
really, really important. But if you sort of take those
fees, it's slow, it's coppers, it's delicate balance, There's a
lot of big vendors, it's integrated system.
Your job is getting out of the way.
The job is to be taken for granted.
The job is to be in. This job is to make the best
(10:57):
workflow tool that specialists in that workflow.
A door because you understand every small click, every angle,
every shadow, every shade, everysecond you've cut down to the
bare metal of it because they didn't didn't come here for you.
They came here to do work and your job is to get out of the
way. But if all of these things feel
super meaningful, this last thing I would would, would offer
(11:19):
everybody to to to trust is thatyes, you're seeing fantastic
companies who are scaling super fast in the application layer,
the workflow layer, and they're building amazing businesses, but
there's plenty of room. There is so much work that isn't
solved yet and there is so much work going on that is not close
heart to the the Hippocratic Oath that somebody took to come
and treat patients, help patients, be more for patients.
(11:40):
And that still has to be tackled.
So our, what we're seeing and is, is a, is a massive explosion
of more and more specialized healthcare tools and they're
beating many of the more generaltools because they can be so
much better to work. So there's so much more that
needs to be built. This is so interesting because
it aligns with stuff I've heard you say before.
This idea that the best tech andHealthcare is trusted but
(12:03):
invisible, That, you know, it addresses these multitude of use
cases we're talking about. And yet, instead of being, you
know, in your face as one of thelatest innovations, as I think
many of us are used to, it is purpose built and specialized
and quiet. How can AI tools like Cordy help
(12:25):
to shift that perception? Where there's a lot of alpha
here is a lot about making sure that the last mile is very
valuable. So obviously all these AIS
allows for lower intra barriers and we're seeing some of the
healthcare explosion. We're seeing other markets too.
There's more, I'm sure CRMS thanever before, and many of them
are sure, I'm assuming or being built on AI instead of classic
(12:46):
database structures as they usedto be building today.
That's not different here. It's the same thing in
healthcare. But the real winners here are
getting very, very deep in the work flows.
And I think something that people don't see a lot is
they're actually growing like weeds.
So we have, we're big fans of a Swedish company called Tandem.
There's a Australian company called Heidi Health as well.
These companies are awesome companies and they're growing
(13:07):
like weeds. That's because they really,
really understand their users and they're making all these
decisions to make sure they get them.
And I think that's sort of important point and it's that
allows them to be much more likeAPLG bottom UPS motion.
And I think that's the fear for a lot of people thinking about
doing a healthcare startup. They're thinking, oh, damn, did
I need to do enterprise sales being all these meetings like go
(13:27):
go to a Minnesota and partner upwith some arcade company and all
that stuff. Actually, you can build a
really, really good company. Bottoms up in healthcare today.
PLG works, pickup is real. Doctors are buyers.
They are take forward. They are not the laggers you
think and there are companies around the globe and every
specialty that's growing like weeds right now.
We have a fantastic company we work with Germany called Nelly.
(13:49):
Nelly is one of the coolest company I've seen that's really
owning sort of the dental, dental revenue cycle management
part and they're growing like it's crazy growth as we're
seeing. So I think it's sort of under
the radar a little bit. I think some investors know, but
I think generally people think that still Healthcare is a
smaller live enterprise sales boring, slow.
It doesn't have to be, but it's obviously if if you want to get
(14:09):
the price for it you need to. You're probably stuck here for
more than than average fun cyclein BC universe.
It's so interesting because there is this idea or maybe
misconception that I think is atodds with a lot of tech
companies addressing the healthcare market, where many of
them think more about how, how do I automate Dr. time out?
How do I try to cut down the number of, you know, doctors and
(14:33):
nurses that we need by cutting away these tasks?
And a, that's not going to land super well with these extremely
overworked and very busy medicalprofessionals.
But also, it often misses the mark of what clinicians actually
want to need in order to addressthe needs of their patients and
(14:53):
to address their own workload needs.
So it seems like there are many builders that are
underestimating how little interests many clinicians may
have in AI itself. They, as you point out, just
want software to get out of the way and work for them.
They want to solve a problem like many, many users do.
(15:14):
But often, I think, as we as builders assume, oh, you really
want to get into the details of this?
No, they, we want to solve theirproblem.
They want their problem solved. So how do you bridge this gap,
especially when AI needs to workmore like electricity, seamless,
embedded and invisible? Yes.
So, so our, our job is we're, we're selling AI infrastructure
(15:35):
and models to people who want tobuild healthcare applications
with AI inside, but they want tobe using purpose build stuff
that isn't rate limited, that has high accuracy, that knows
the medical terminology, who's that's billed for embedding into
these systems, that has unique endpoints and SDKS for revenue
cycle management or ambient documentation, all the things
going on in healthcare. And that's where we come into,
(15:56):
we want to be that infrastructure and, and do it at
a price point that's super, super competitive with much less
accurate general purpose models.That's what we hope a lot of
people start building on openingeye or whatever.
And opening eye is a fantastic company.
They'll build fantastic things, But at some point they might
want to get into parts of the workflow, the problems that are
daunting healthcare specific. And there we feel we have a real
opportunity to help. And I think it's, it's let me
(16:20):
down such a rabbit hole as sort of a builder wanting to build
for builders, meeting so many ofour customers customers.
So I had a case not that long ago working with some of working
with this case, one of the coolest, I think fast radiology
companies in the world. So they're doing radiology
equipment and software and we'vemet a lot of their customers.
So actual radiologists, I obviously know what radiology is
(16:41):
before coming there, but I've never like been in, in the room
where they do their work right? And it was amazing work.
And I had heard so many startup pitches in my time, both in the
Oregon and seven in Europe. Like all we're automating away
for like radiology reporting. There's no radiologist anymore.
They don't have a job. I've seen that title so many
places. It's just so funny getting into.
Then you're seeing a person and you've been told by media and
(17:02):
startup pitches like no radiology radiologists wants to
like, examine these images. They just want to get AI to do
it and all that stuff. That's not my experience.
My experience is that we have, in this case, a school of
radiologists who are like Sherlock Holmes.
They can glance at something andthen they go full minority
report on it. They can find small, small,
small Nuggets of gold in there. And they do it like decision
(17:23):
machines all day and they enjoy it.
They're good at it. Could they get AI to help them?
Is there enough radiologists? They could definitely get
somebody to help them. And yeah, there's definitely not
enough radiologists and AI couldpluck some holes.
But the point is, I think the narrative always becomes this
like a plug and replace. I think that sort of the big
missed opportunity both sort of from a capital allocation
problem and from sort of a startup problem, is that if we
(17:46):
really went out there and we're allowed in the room and see how
all this works, the job is a people's job.
It's healthcare. It's not like health task or
health solve. It's healthcare because care is
sort of the continuum you need to do it and a lot of people
working, they actually enjoy it,unlike many people in Silicon
Valley thinks like it's like it's so like non rewarding to be
in healthcare because they've read all the bylines in in in
(18:07):
the COVID coverage. True, COVID was probably not
very rewarding for many because the system was so rigged against
so many providers. But I think a lot of people
still go to healthcare. I have a lot in my family, at
least they went there to have impact, to do care, to make
change, to spend time with patients.
I think an example like this would be all just, again, I know
they're not exactly like the, the, the professionals with the
(18:28):
most patient time, but they're really good at what they do.
So the job isn't here coming in saying you would like you
trained 17 years to be the world's best decision machine,
Sherlock Holmes here, and you'renot doing like you shouldn't be
doing it just be automated away.The job is to make sure that we
can make them be even more productive doing it.
And that might be AI doing some,they do some.
It might be AI augmented, it might be AI doing all the
reporting. That's what we think, but
(18:50):
ultimately it's about listening to where the market is going and
how to deliver more care. It's not to sort of deliver more
continue. And I think in this case, it
actually impacts the, the, the, the capital allocation question.
So the the Silicon Valley narrative as well, because if we
think about it, money will just flow where they get the most
ROI. And obviously great ROI would be
AI does all of it. But in the in get again,
(19:12):
Healthcare is still a care continuum.
So if there's a lot of people doing care, but some of them are
augmented by healthcare, they'remore productive.
That means they have better around a time.
That means more money will flow to these departments that are
more AI enabled, which in turn will mean they can hire more
people than if they can get themas prolific in AI, they will
create more outcomes, more people will flow there.
There'll be more money, more opportunity, better care, better
pay. And all of a sudden the best
(19:34):
people will be working there. And those organizations where AI
enabled will grow. They won't shrink away to like
some weird like dystopian futurewhere there's no care at all.
I love this example of radiologybecause there's such a high
impact specialty and it's clear that adoption is occurring in
this area. Let's unpack this a bit more.
(19:54):
Can you tell us more for our audience?
You may, you may not be aware what exactly is AI doing to
enable radiologist today and andwhere do you see the future?
You mentioned admin work. What else?
So, so I can at least talk aboutthe cases we're heavily involved
in. And I, we are not specialists in
radiology. We have people here who work
(20:14):
here who really know the, the space, but our customers are
world class. And what they come to us saying
is that there's actually a host of products that are sort of
last generation AIS. So classic maybe supervised
machine learning or maybe even just like brute force models
that did stop like dictation. And imagine you're sitting there
looking at the screen at the radio.
It's like say field #2 put that stuff and it's an 8B AT and
(20:36):
there is a this kind of fractureand you're talking medical
language, but you're also talking to an interface.
You're telling denoting where mouse go, where to click, what
to do. Obviously that's not what the
future should look like. You shouldn't be like worried
whether or not it got your commathe right or whether or not it
moved the cursor. Yeah, it should be cognizant
enough to know that you're operating in the system and it
(20:58):
should be agentically enabled enough to go and plug all the
information you're just saying to it in healthcare language as
fast as you wanted, in dialects you wanted, translating into
language like you wanted into all these interfaces and start
all these downstream tasks. That's what we're thinking a lot
about, we call natural language dictation, which is just having
a blast with an LLM copilot thatis AI enabled and agentically
(21:19):
enabled to go inside all these radiology platforms and help
start finish workloads. So you're not just saying move
cursor to point B, you're actually just talking.
And then you have something thatfeels cognizant that isn't sort
of an after the facts, like few shot, one shot LLM writing a
summary that's riddled with mistakes.
It's something that happens in real time.
That is sort of a recursive reasoning and making sure that
(21:41):
it actually finds the facts, understands the facts and starts
the actions needed. What's a more productive
narrative to leverage here as AIcompanies think about adoption
then within emerging healthcare verticals?
Obviously understanding your vertical, really getting deep in
(22:02):
the language of it and not trying to sort of, I've heard so
many health healthcare professionals talk about sort of
the, the health like the tech savior complex.
It's like, don't land with the savior complex of you solving
anything. Like we all humbly need to know
that these are like, these are people who usually chose their
career path because they're among the smartest of us and
we're just there to serve them and make that happen.
(22:23):
So I think the narrative is, wait, wait here.
It's just to make sure we get away from the dance floor and
make sure, like, the food is served and the light is on and
everything works. And the people are really good
at those workflows. They are the people we see
growing like weeds. Enablement, not replacement.
Absolutely. This reminds me of the
conversation around AI and different creatives where yes, I
(22:46):
in that instance, there is I think a side narrative of, oh,
we're going to enable everyone to, you know, do more creatively
by giving them AI tools. But often expert artists don't
really want AI to paint for them.
They enjoy the act of creation. They maybe want to be enabled in
areas that are around their key tool sets.
(23:07):
Is this a fair parallel do you think?
I think it has a lot of overlaps, that's for sure.
And I think the respect for for the work is key.
And I think that's very much sort of built into some of the
critique. And so if it's some of the the
sort of the phenomenal headlinesof creative sort of going away,
I think Mira Muradi at some point said when she was at Open
the Eye that they will replace alot of creative work.
(23:29):
And some of those creative jobs might never, should never have
been there. I'm sorry, Mira, if that's not
correctly quoted, but I think the point was just obviously
there's a building critique herethat like if it can be replaced,
maybe it should have been in theget go.
I just do think, especially for creatives, like I have never
seen a chess match between 2A ISI have no plans of seeing like
a Dota match between two Dota playing AISI have known, but a
(23:51):
Dota like world class team of five playing a tech team of five
AISI might TuneIn that might be fun.
So I think this is a lot about the interplay.
It's not about the pureplay. AI is going to be cool and a
Genting AI in Healthcare is going to do so much work that
nobody signed up to do. I promise you, very few people
in the world find like billing, coding, fantastic endeavor.
(24:13):
Some do and they're amazing and we need them.
We'll always need some, but they're rare.
They're rare and they should be rare because it's hard work and
it's complicated. And maybe those people who are
doing it today who don't love doing it, they could be doing
patient facing stuff instead. But that's I think the
opportunity of healthcare. Yes, we'll replace some work,
but for the majority of it is way more about open ending or
enabling. I like this idea of
understanding the intrinsic motivations of professionals and
(24:37):
people in order to tailor what we're building to them.
And I think it speaks to being customer obsessed and being user
obsessed in a way that too oftenfolks who come in with a savior
complex, as you put it, or maybearen't listening to the people
on the ground, but instead are saying, oh, here's how I can re
(24:59):
envision this market. And while it's great to have a
vision for the market and there's significant
opportunities to enable physicians in particular, 24% of
whom think about leaving their role weekly due to overwhelming
workloads. You know, clearly there's a huge
opportunity there. But failing to talk to them and
understand their pain points is just a huge mistake.
(25:22):
And it is very clear to me that you're not doing that.
And you're both addressing the needs of your, your own clients,
the builders who are enabling healthcare providers and then
also digging into the individuals within the system.
Yet there is this significant workforce challenge.
You know, we, we mentioned there's challenges of the admin
work, there's challenges with what I've heard you refer to as
(25:45):
pajama time and the lack thereofof, of folks doing notes at
night. And there's a gap in potentially
10 million trained healthcare professionals that will need by
20-30 that likely can't be filled through traditional means
alone. So what needs to fundamentally
change in our approach to AI development and deployment for
it to truly address both the looming healthcare professional
(26:10):
shortage? I mean, frankly, there is one
already and also the pains that individual providers and
physicians are feeling as they navigate this challenging high
stakes regulated environment that's extremely high stress.
It's such a good point and I think The Who actually upgraded
their number of the the caveat or the delta of of the
(26:32):
healthcare professional will need by 20-30.
It's hovering between 10 million.
We lack just to deliver the carewe delivered today, not the the
cooler version of healthcare we'll have learned to do by
then, but just to do what we do today.
It's at least 10 million we'd like globally healthcare
professionals. So you're right, there's just a
problem if you can't do what we do today.
But if we look backwards and we assume it's the same curve of
innovation and novelty, we probably don't want to do the
(26:54):
healthcare we do today in 2030. We probably want to do way more
preventative care, longevity, all that stuff, right?
So you're right, there's a problem and I think sort of one
of the things we're seeing sort of what's the sort of the low
hanging fruits like in energy weall want like maybe some of us
wonder fusion or some of us wantfission, some of what want
something like solar. But some things are like easier
to do like solar and something are harder like fusion.
(27:16):
And I think in our case in healthcare, if we just look at
the 25% of the the workflow, so I said before that we have sort
of 40% issue of work done is like language based.
If we look at sort of workflows that are language based in
healthcare that are ripe for AI innovation and we just take the
25% most ripe and we add AI to the parts of it that could be
automated. I think it's McKenzie or
(27:38):
etcetera at McKenzie. That said, if we do that, we can
actually move $1 trillion worth of HCP salaries away from admin
that's in workflow like collecting data, process data.
That's like the, the, the catch all kind of workflows, right?
If we did AI optimize those, we have up to a trillion dollars.
(27:58):
We can move back into care or somewhere else, but back into
care seems nice, especially if we lack 10 million people.
And if you look at sort of the what does like 1 trillion,
obviously it's very different salary curve in, in parts of
Europe versus US and so on. But if we sort of look at the
the the gap here is actually possible to map quite a bit of
the most critical ones by just doing the most optimal, most
(28:21):
obvious reallocation of time by using AI healthcare.
Let's drill down on solving thisproblem because you've talked
earlier in our conversation about this major opportunity
around, you know, 4 out of 10 hours in healthcare are spent on
language tasks. So this seems directly
applicable to AI as it stands today.
(28:44):
And then, I mean, the scope of this is staggering, obviously,
you know, potentially moving a trillion hours back in the
workforce or, or freeing up those hours.
How would you break down how this would all work?
We need a Cambrian explosion. We need so many more
opportunities for healthcare professionals to leverage these
(29:08):
kind of technologies. Today, we have a lot of
healthcare still running on premise.
I think that's sadly the truth. If we look at the big, big, big
EHR platforms like Epic, a lot of it is still on premise.
We're seeing it move more and more to the cloud.
And I think Microsoft and Amazonand many other cool companies
are playing a large role in making that happen.
So we will unlock more and more opportunity to add more of these
types of technology as healthcare moves to cloud.
(29:31):
Secondly, we need to make sure that we find better ways of
enticing the big sort of systemsof record to start opening up
their doors. Some companies like Epic are
like they're doing all sorts of initiatives now to open up the
doors to more AI vendors. They're writing AP is they're
opening up. And I think that's the path
forward. Some of our customers like
(29:52):
Dataloose, the world's first biggest DHR platform, they cover
patients from Bolivia to Germany.
Like it's a massive undertaking.And they have a massive tech
team that are doing a lot of this stuff to enable more and
more tools to plug in. So more cloud, more plug in,
more tools, more builders that dare specialize.
That in turn I think will be theCambrian exclusion.
And if we can then get some AI that doesn't hallucinate as
(30:13):
much. I think the healthcare
professionals actually want to adopt it, even in fax riddled
Germany. How are you addressing the
hallucination problem? What's your approach to
evaluations? I wish that was the only thing
that relaxed to make sure we we we have opportunity to
hallucinate less. The the boring answer is a lot
of tedious work across the factory 4.
(30:33):
So if we think of ourselves as afactory 4, we input health data,
we output health data or health reason.
No. And, and in that factory 4, we
have many models, many machines.And at all times we need to 1st
off, we need to benchmark. Was it the quality of all the
data we got in? Did we understand it?
Did it, did it translate well? If we transcribe it, did we
transcribe it? Well in the factory, every
(30:54):
component that's interacting hasto have benchmarks.
So we need to understand, removedata.
What happens? Where does it happen?
We also need to understand what's the contribution of this
machine in the factory 4, so we could plug it and play it.
Oh, somebody built some really cool shit over here.
We're really inspired by this company.
How do we learn from finally, when we output it, we always
have public benchmarks and public benchmarks to be honest
(31:14):
is for the vast majority of not very useful.
If you go to the Hugging Face and find some of the big sort of
medical benchmarks, pub Med, there's many others.
You look at the the leader, the scoreboards of many of these
health data sets. The leaders are non commercial
models. I think everybody in machine
learning will have a good guess at what that means.
It's probably because it's not very useful in practice and
(31:36):
there's probably really good reasons to not bend and hammer
on anyone. But there's probably really good
reasons why it was trained in a way that makes it really good at
tests, but not very good in the wild.
So I think it's not just benchmarks.
It's an entire infrastructure, it's many pipelines.
It's understanding like not all data is created equal, not all
data processing is created equal, and every component in
managing this ecosystem of reasoning is a massive
(31:59):
undertaking because we need to be able to actually explain what
we do, why we did it, why we changed it when we did what's in
the model cars, like what's in the food we're serving, what's
the ingredients we used, why didwe use them, when did we move
them? How do we launch it, where is it
hosted, at what level and what state are we encrypting?
There is so many parts of this puzzle that it's to to many
(32:20):
dismay if they want to play in the infrastructure, land,
healthcare, it's sadly not just one thing and there's not just
one benchmark. It is an entire infrastructure
of solutions that has to come together.
I think that's a big learning from RCA is like you can win
healthcare by being good at one thing.
You have to be good at many things at least quite a bit, and
then you can be OK at some. Then you can decide to not be
good at all at some others. Like us.
(32:40):
We're not very good at the last morale, the workflows.
We have partners who are fantastic at that part, but we
want to be really good at infrastructure part.
I think it's wonderful that you've found the focus of where
you want to approach things. And I appreciate the the detail
that you're bringing to both explain ability and
(33:02):
observability as well as customization.
Because too often I think we areseeing approaches in different
verticals with AI where you justsay, I'll throw a big
generalized model at it. It's going to work great.
And in fact, you need to really customize the metrics you're
(33:22):
using, the benchmarks you're applying, how you approach the
task itself. You know, what tools are you
calling within an agentic system?
Everything needs deep customization and infrastructure
work to effectively address problems in in challenging
industries, particularly for highly regulated ones.
You know, when you introduce nondeterminism, there is an added
(33:44):
element of of risk, which an added opportunity to but there's
there's so much to solve there. So you know, I'll say a Galileo.
Obviously we are a huge fans of this approach for customizing
benchmarks and metrics and really building hybridized
systems that include both that like expert feedback, but also,
(34:04):
you know, leveraging LMS or SLMSlike our Luna models of judges.
So I so I love that Cory's taking the same thought process
of, OK, how do we make this explainable for our customers
and for our users? Because without that, it's hard
to really deliver value, particularly in these very
consequential workflows. And then how do we customize
(34:26):
this to really meet their needs instead of just saying, Oh yes,
here's this generalized benchmark that we can apply.
So love that you're thinking that way.
I think that's absolutely the right approach, but I also
imagine there are major hurdles that you're facing, whether
they're technical, regulatory, cultural, to achieving the level
of trusted automation in healthcare that I hear you
(34:46):
seeking to achieve. How does Cory take a compliance
first approach to help address these?
Do you know what's really helpful?
Being from Europe, we're really good at compliance.
We're really good at it. So if I were doing AI for for
many other things in the world, I wouldn't be saying this.
It would be an inhibitor. But we actually have a really
(35:08):
cool, very, very, very large US customers who are launching very
soon. I don't exactly know when this
is going live, so I won't name them yet.
They came to us. They had a very, very long
purchasing site. They're a massive company.
And one of the things they like was like, OK, so you guys are on
like AI Act and GPR and all thisstuff.
And they're actually moving towards buying from AI vendors
who are more in healthcare, in healthcare.
This is healthcare aligned towards stuff like AI Act
(35:30):
because they think no matter what the market will drift
towards some of this stuff because yeah, some of it is like
red tape, we should remove it. But there's also a nucleus of
truth back to you want to be able to trust it even in the
high stress environment. And I think yes, there is a lot
of red tape in Europe. That's not great, we all know,
but there's a lot of red tape that really makes sense and a
(35:50):
lot of furrow People really thought about it.
And I think that's something that coming from Europe, having
built a company grounds up here,like our first many, many models
we like we pre trained, we did all of ourselves.
So we really, really know how tomake a good soup.
And I think that every chef's knows if you can make a good
soup of simple ingredients, you're on to something.
But also allows us to really control went into the models.
How do we train them? How do we like structure it?
(36:11):
What do we do post training? If you actually do the work,
it's much easier to give somebody else a recipe.
I definitely agree with you about the ability when you have
already addressed certain regulations and compliance needs
to then apply that approach. Being able to have that baked in
from the ground up is so important.
And I'll say on our end, we've seen this in financial services
(36:33):
and banking already where, you know, because we're already
working with JP Morgan Chase, because we're already working
with Citi, it becomes so much easier for us to have these
other conversations with major banking and financial service
providers because we get their regulatory concerns.
We know the customization needs they have.
And I'm sure that's the case foryou across Europe, across the
(36:55):
world, as you are able to apply this regulatory framework that
you've you've already understoodand become compliant with.
I'm going to give you a leg up in consequential industries.
I think we see this also with, you know, AI defence and
elsewhere, but how do you get AIto a point where it genuinely is
(37:15):
reducing workload? Like, great, we're, we're
complying with regulations, we're, you know, handling the
different situations we need to,but how do we make sure it's
invisible and reliable? There's so much infrastructure
work that goes in behind that. What are you working on at Cordy
to support this mission? Thanks for asking this.
(37:36):
This is where you can keep me all night.
An example is in healthcare right now, everybody's pretty
hyped about AI Scribe. So these like ambient tools,
you're in sort of outpatient care.
You go to your doctor and they turn on a mic or their phone and
they start recording. And then there's another them
listening, recording. And then when it's done, it sort
of spits it into this data pipeline.
It transcribes, then song rises,then you blurb out some text and
(37:58):
here you go. Your note is done.
And always today these like notetakers are like they're saving
so much time. Yes.
But but we, we did a survey of 2000 of the earliest adopters of
these cool tools in the US and they told us on average, they
spend at least three hours a week just like betting and
correcting their AI. And they say it's a constant
worry for them whether or not the AI had like set, shared or
(38:19):
done something they like had miscollect, mis recollected or,
or didn't change. So we're creating workflows here
that shouldn't be there, right? That's not the promise of AI is
that like we'll give these fantastic HTPS these new tools
and then they'll sit and get worried about those tools like
they're burning out. We should not get them more
worried, right. So yeah, they're great.
We fought like that's exactly the kind of thing where like an
(38:39):
infrastructure company like us built for it can make a change.
So so we we offered up somethingcalled facts are.
So it's a it's a recursive reasoning pipeline.
So it's not just sort of an LLM that one shots or few set
documentation, but it actually instead listens in.
So you turn on the microphone. You've maybe before you build in
some like general purpose LLM, alot of model whatever.
Now you have Corey instead of just waiting and like batch
(39:02):
processing a few shutting it, we'll listen in all the time.
So you open it. And while you're talking to the
patient, we are doing recursively revisiting with
using like an orchestra of of anelims.
What is the sort of saline factsthat are really important and
can we revisit their importance as the conversation changes?
We all know, like we've been having a conversation like this
(39:22):
or we're having a conversation over wine and subjects change.
But in a transactional conversation, like in
healthcare, you're transacting where to go, where to stay on
medication, to do. There is a kernel of truth that
needs to be respected. And all of the side time we
visited. But if you just do batch
processing, these little limbs will spit out shit.
They will anchor on stuff. They will have like indexed on
something. It will maybe even hallucinate.
(39:45):
We're all the time trying to make sure that there's an
orchestra and there is a series of models as judges passing
judgment on whether or not this reasoning pipeline gets it.
And what we're seeing when we benchmark it on on some of the
big public benchmarks is that not only will we find 49% more
of what matters, the really important healthcare
information, we will also reducelike the verbi verbose sort of
(40:07):
classic. We all know in the limbs can be
pretty verbose, right? So we can reduce 88% of what
doctors would deem verbose or noise in these documentation.
And if you actually try to test whether or not they like the
doctor, both find it like accurate, concise and complete,
complete being. This was something I would have
written. It's like up to my standards.
When they work together with a real time flow, like recursive
(40:28):
reasoning from our effects, our model they actually feel is as
complete or close to as what they did, which is a vast
difference when there's a limb spat out something that didn't
feel theirs. So this is the kind of thing
where we can make real decisionson the infra layer that makes
the like the the the entire experience for our customers.
Customers completely different without it, like at the end user
(40:49):
stage feeling much different. Let's dig into this infra layer
a little more. I'm curious to understand more
about how you're applying judgesto double check outcomes and
evaluate what. What type of judges are you
using? You mentioned using multiple
judges. How are you waiting them?
So some of it that we actually have in our documentation right
now, which is live on, on on ourwebsite, you can go build it.
(41:11):
And I think some of our customers might have an impact
on how we want to do it in some of our enterprise deals as well.
Obviously, the challenge is to find a scalable way of
understanding how to pass judgment on something that is a
moving target, which is hard. So we did, we do need a lot of
data sets. We need a lot of holdouts, which
luckily we've been here for almost a decade.
So we have done a lot of that. And then it's about really
understanding the workflow it's embedded into to make sure we
(41:33):
applied in the right point at the right place.
Furthermore, at the end of it, we're lucky in our workflow that
a user has to pass judgment, right?
So a doctor has to say what you did was cool.
So we actually, at the end get aquite cool golden label of
whether or not it's as complete or as concise as somebody would
actually use. It regular human feedback to use
for reinforcement learner exactly and.
And what we even see is that when you normally get these like
(41:56):
few shots and you sit in the endwith this like build wall of
text, like the, the quality and the feedback is obviously
temporarily it's all batched, right?
Whereas if you're all the time, you can just like, instead of
sitting there at night pyjama time writing and rewriting it as
you recall it, we just like glance your screen and you can
like click, click. Then you've given it feedback
(42:16):
that actually adds A temporal element as well that we can
stand and use. And all of the sudden we can
stream the entire conversation for the API, understand as well
the interaction with the AI as it moves.
And all of the sudden we don't just have a really good golden
label of concise and completeness.
We also have an understanding oflike when you interacted with
the API for your platform, what happened?
What did you do it? What did you interact with?
(42:37):
What did you change? So all these things helps a ton
looking. Ahead, as AI models become more
capable, where do you see the next significant opportunities
to move beyond administrative tasks and begin to safely and
effectively support diagnostics,decision making?
And ultimately not only drive improvement and freeing up time
(43:02):
of the clinicians, but also directly improving patient
outcomes in a more, you know, regulated and compliant way
while still respecting the clinician's role.
I think it's a lot about sort ofif we took by to it, there is
something that's like new workflow, new market.
And in that like new workflow, new market, there is stuff today
(43:23):
that hasn't happened in healthcare yet because we might
not afford it, It might not be obvious that should be happening
in the future. And I think there's some cool
companies out there trying to map that space right now.
Some of it is like like existingworkflow, existing market that
might be like adding AI to automate a ways sort of
secretaries. I think a much more interesting
workflow is like how many conversations could you have?
Or me for that matter. Let's say I picked up a new
(43:44):
medication. I have never taken it before.
Hopefully I'll never do it again, but that means I have 0
contextual there. My physicians usually don't have
time to call me and select. Tell me more about that.
Like that like weird feeling of like airiness you get in your
head. Like it's a soft signal.
It's not probably not a leading indicator.
It will go away so nobody will call you, but if you're the
(44:04):
average cost per minute of high quality medical reasoning with
solutions and infrastructure like us can be like orders of
magnitude cheaper. Why wouldn't you push amazing a
is at calling all your patients?If that's in pharm eyes, that's
in like physical rehab. It's like psychology,
psychiatry, primary care, outpatient care, you name it
home care. Why would you spend way more
(44:26):
time just reasoning with the patient, cautioning them,
helping them, guiding them if you could and you actually had
an infrastructure that was safe enough to not to lose any
weight, take all the pills, it'sgoing to be great, right?
We all heard the like Google like offering patients to eat
one rock a day because that's like a fictive study from
Berkeley has said that's great, right?
If we can control it, then like all of the sudden bring compute
(44:49):
costs down will be such an unlock for like you work for the
new markets that we haven't thought about yet.
Andreas, thank you so much for staying up late in Copenhagen
today and joining on the podcast.
It's been such a pleasure havingyou on the show.
Where can our listeners go to find out more about Cordy and to
learn more about your work? Hey, thanks and thanks for
having me. I really enjoyed the the the
conversation. We're at Cordy dot AI.
(45:12):
Go sign up, ping me on Twitter or LinkedIn.
My handle is Andreas Cleave or ping us at Cordy.
Do it on Instagram. We'll, we'll shoot you over some
credits. You can start building if you
have a cool use case where you need something bespoke,
something hard. If you're a University Hospital
wanting to do real clinical research with an AI partner.
We're, we're not sort of the latest fad.
(45:33):
We're here to stay and we'd loveto build cool shit with you.
I love it and I honestly have super enjoyed this conversation.
It's been such an interesting look into this.
I think misunderstood vertical as you put it earlier in our
conversation. There are 1000 different
verticals within healthcare and they all deserve their own
(45:54):
treatment. There's such an opportunity
here, and it's inspiring to hearthe stories of companies like
Cordy that are focusing on highly regulated consequential
verticals and making deep inroads by being considered
about their approach. So tons of great lessons that I
think our listeners can take away from the approach of
QWERTY. And hopefully a few folks on the
(46:18):
healthcare side of things will be reaching out to you after
this. So we'll certainly link
everything that we've discussed today in the show notes.
Super excited to learn more about your evaluations and how
you continue to build out the QWERTY platform.
There's clearly such an opportunity here.
And for our listeners, if you enjoyed this episode, support
the show by subscribing. Give us a like a comment, a
review. It matters a ton.
(46:41):
Take a little pajama time to take two or three minutes and
just, you know, drop that comment.
Those drive the algorithm in ourfavor, whether it's on LinkedIn,
whether it's on Spotify, YouTube, wherever else.
We deeply appreciate your support.
Thanks again for tuning in. And Andreas, thank you so much
for joining us. It's been a pleasure.
Hey, it's been a pleasure and good good work on all fronts.
Big fan of yours. Thank you very much.
(47:01):
Appreciate it.
I think the whole sort of animosity, like healthcare
doesn't like technology. Oh, doctors are are take
lackers. They don't get it.
I just think it's like super boring.
It's a little bit like telling NVIDIA 10 years in, like this
whole like GPU thing, like you've been pretty slow at it so
far. Is it really worth doing?
We are back on chain of I thought I am your host Connor
(00:22):
Bronson and today I'm delighted to have Andreas Cleeve, CEO and
Co founder of Corti with me. Andreas, welcome to the show.
Great to see you. Hey thanks for having me.
Today we're diving into a sectorwhere AI promises immense
transformation and where Corti are already replacing single use
apps with medical grade AI. That's right, we're talking
(00:42):
about healthcare. There's unique challenges to
actually using AI within healthcare, but there are
amazing opportunities. The field is often cited as a
high stakes environment where the margin for error is
virtually non existent. We're dealing with people's
lives. How do we responsibly and
effectively integrate AI into such a critical domain?
(01:07):
We often hear about healthcare as this scary place for AI, so
I'm excited to unpack everythingtoday with you, Andreas.
To start, can you give us a brief overview of where AI
adoption is currently here in June of 2025 in healthcare?
Yeah, it's a great place to start.
So healthcare, as most of laypeople, me included, knows of
(01:30):
it is, is riddled with language tasks.
So roughly 40% of salary dollar spend to spend on work that
relates to language. So obviously talking to
patients, the part we love all, especially us patients, but also
writing notes, going to HRS, writing letters, writing quotes,
invoicing. It is all the, the, the lifespan
of healthcare. There's a lot of language and
(01:51):
text tasks. And obviously we have a new
thing in the arsenal with all these the transformer based
models that are very, very decent at, at doing language
tasks. So what we're we're we're seeing
is a lot of adoption in how can we take some of the what like
the cold hands based language jobs.
That's like tapping away your keyboard when you should really
(02:12):
be be focusing on the patient and trying to see how we can get
these models to to do some of that work.
It's so interesting because I think my perception at least is
that there are barriers, regulatory or otherwise, to tech
adoption in healthcare. We may see surgical equipment
get adopted rapidly in some cases, but often there are
(02:36):
technologies that are coming outof Silicon Valley, for example,
that maybe take longer to actually get adopted.
Are we seeing that with AI, particularly with the capability
of general purpose models and the opportunity for them to, you
know, transform all these 40% oftasks you mentioned, how can
(02:57):
they reasonably meet clinical and compliance standards?
Yeah, it's, it's a, it's a fair question.
I'm probably going to rant now, but I, but I think I, I, having
lived in SF as well, I think this whole narrative about
healthcare being bad at adoptingtechnology, it's just really
boring. And it's a Riddle with lack of
ambition. And I mean, I get it if you have
(03:18):
you used to build like grade like SAS tools for your
marketing team, who as part of their work spending all their,
the computer and they're adopting stuff fast, coming to
healthcare feels super slow. And I hear so many people be
like, oh, that, that law sort ofruins this as an investable
class and opportunity. And it's like they won't adopt.
But if you, if you sort of stop for a moment and think about
healthcare, you go to your hospital, your clinic,
(03:39):
especially your hospital, but golook around.
How much technology do you see? I sit here at our offices in
Copenhagen where we have what, 70 researchers doing machine
learning and I would bet you that per person, there's more
devices in the in the healthcareinstitution in per per room, per
clinic than there is here. Totally.
So they adopt tons of technologyand they do it quite fast, but
they have a pretty high bar for doing it.
(04:00):
So I think the whole sort of animosity, like healthcare
doesn't like technology. Oh, doctors are or take lackers,
they don't get it. I just think it's like super
boring. It's a little bit like telling
NVIDIA 10 years in, like this whole like GPU thing, like
you've been pretty slow at it sofar.
Is it really worth doing? I think there's just such an
opportunity building something like life's worth worth that is
(04:22):
really compounding every year that is massive because
technology is just, it's a different game here.
It's a game about being taken for granted.
So like if you're a lying on your bed and you're a patient,
like, and somebody comes in and say, hey, we're going to do an
MRI scan. Like do you think it would be
cool? I'm sure you wouldn't.
So it's very facetious rhetorical question, but
obviously you wouldn't be like, Hey, I want to see your ISO
(04:42):
standards first. So in a universe where you
actually want to take for granted that everybody is up to
their absolute best, they're delivering the best work they
can do. There's the lives work.
Every time they meet you, they are doing the best thing with
the best tools that are most applied to to your case and
they've been tested into oblivion.
If that all has to be true, thenwe can't also, while we're
(05:04):
sorting like kombucha mode and we don't feel any pain, be all
worried about oh, like healthcare, will it ever change?
Healthcare doesn't necessarily need to change.
It needs help. And that's, I think sort of one
of sort of the hidden secrets here is like Healthcare is only
for a, for a, for a boring, stubborn people who actually
want to have impact and want to be taken for granted.
I love the passion you bring to this because you're completely
(05:28):
right, there is a, a lack of ambition as you put it, where
often we have another B to B sass product then oh, I can't
get into healthcare because it'snot needed there or it's, it's
not a great fit or simply because they don't want to deal
with personal identifying information and healthcare
standards around that. And yet there is so much
(05:51):
technology in healthcare today. And I know you have some great
statistics about where the rubber meets the road here.
How many physicians, or anyone for that matter, are using AI on
a daily basis in the medical field?
So what are the real world outcomes when using purpose
built models like Cordy? Yeah.
(06:13):
So, so when the first generationof AI models build on us are
general purpose stuff, we're seeing a lot of pick up in
certain parts of healthcare. Now we get a little bit tedious
and boring here, but a lot of people think about healthcare
this big sort of monolith. It's definitely not.
And then they look at some of these startups and some of them
are amazing companies like, well, one of my favorites is a
Bridge American fantastic company who offers AI scribe and
(06:36):
AI assistant and copilot technology.
They've built an entire platformto to do some of the best
physician augmentation. They're seeing like any SAS
founder would be envious of their their kind of growth.
And they're like really, really good at outpatient care.
There's also really cool companies really good like rad
AI American company doing a documentation augmentation in
(06:56):
radiology. They're a super cool company.
They're really deep in inpatientradiology.
I guess some outpatient as well,but they're focused elsewhere.
We have a lot of like we just saw a fantastic IPO of Hinge.
I'm sure Hinge is using tons of AI.
There's a completely different workflow.
So the point is just the rubber meets the road a little bit
differently everywhere because Healthcare is not just one
monolith trillion dollar admin market, it's 1000 billion dollar
(07:18):
opportunities. So it's a very, very different.
But what we do see a lot is thatin cases where there is a very
large language load to get mandate, like imagine you're my
doctor and you want to make sureyou're sending me somewhere
else. That might be to our pharmacy or
it might be some more treatment to make sure you're doing the
right thing. You have a host of things you
need to take care of insurance. Am I covered or not?
(07:42):
What's the right path? What's the right medical data,
clinical data? Is there any research I need to,
to just double click on and what's the procedures that
usually offered to somebody likeme with the picture of symptoms
I I'm presenting? All of these things has to go
into sort of 1 assessment. And in these cases, usually we
see a lot of notes and a lot of note bloat.
(08:03):
In these cases, AI is really meeting a hungry group of users
who's spending way too much timelate at night.
It was like many called their pyjama time thinking about what
happened at 2:00 PM today and what did I say and what
happened. I had leave me meet a lot of of
healthcare professionals who arein those cases where they did,
they had back-to-back to back all day and I was sitting at
night. Actually, they should be
spending time reading out stories to the kids, but they're
(08:24):
sitting there tapping away at the keyboard.
And those are the kind of cases that's really working, that's
getting picked up where you can get a far, far along away with
with the general purpose models out there.
But we're just scratching the surface though.
Yeah, I love this description ofhealthcare as 1000 billion
dollar opportunities because you're right, so many outsiders.
(08:47):
Do you see healthcare as a monolith despite there being a
ton of differentiation? How would you advise
technologists who are interestedin the space, who maybe are
looking to drive AI adoption, torethink their perception of
healthcare as an industry to address these diverse markets
(09:08):
that you know are more granular across the the entire endoscope
of healthcare? I think the, the first thing
comes down to obviously wanting to do it.
It's not easy work, but that's the entire point.
It's not supposed to be, it's supposed to be really hard, but
it's really rewarding to do whenit's working.
And obviously there's some fantastic companies building
healthcare and health tech and health and being among them is a
(09:31):
very profitable opportunity. So there is both the, the, the,
the, the personal part and there's obviously the, the, the
vision impact part. So First off, we need to be
willing to spend the time. Secondly, I think it's important
to remember that when we sit here and we enjoy spending time
in design and we discuss what new cool design tools can I
augment my feet mouth stuff with.
(09:51):
It's a core part of our designers workflow to think
about it, which tools they're using.
It's not a core part of work fora physician.
Hopefully to think a lot about which part of the plug insurance
from our EHR providers have my bosses, bosses bought, bought.
They should be thinking a ton about that.
Obviously, hopefully they get what they need.
The point is the, the true goal here is not for you to to be
(10:12):
sort of on the dance floor. It's to make sure the people who
are on the dance floor, they have the most great time on the
dance floor. And that's very different from
other SAS products, which, whichare much more sort of high
identity product for the users. So I interviewed tons of
salespeople and many of them arelike very religious about do I
use Salesforce and hotspot? Obviously that happens in
healthcare too, because switching costs are high.
But it's not because I like the physicians we meet take like a
(10:35):
great pride in being an epic person versus being a certain
person. It's because they're good at it.
And I think that that's sort of a second part that that's
really, really important. But if you sort of take those
fees, it's slow, it's coppers, it's delicate balance, There's a
lot of big vendors, it's integrated system.
Your job is getting out of the way.
The job is to be taken for granted.
The job is to be in. This job is to make the best
(10:57):
workflow tool that specialists in that workflow.
A door because you understand every small click, every angle,
every shadow, every shade, everysecond you've cut down to the
bare metal of it because they didn't didn't come here for you.
They came here to do work and your job is to get out of the
way. But if all of these things feel
super meaningful, this last thing I would would, would offer
(11:19):
everybody to to to trust is thatyes, you're seeing fantastic
companies who are scaling super fast in the application layer,
the workflow layer, and they're building amazing businesses, but
there's plenty of room. There is so much work that isn't
solved yet and there is so much work going on that is not close
heart to the the Hippocratic Oath that somebody took to come
and treat patients, help patients, be more for patients.
(11:40):
And that still has to be tackled.
So our, what we're seeing and is, is a, is a massive explosion
of more and more specialized healthcare tools and they're
beating many of the more generaltools because they can be so
much better to work. So there's so much more that
needs to be built. This is so interesting because
it aligns with stuff I've heard you say before.
This idea that the best tech andHealthcare is trusted but
(12:03):
invisible, That, you know, it addresses these multitude of use
cases we're talking about. And yet, instead of being, you
know, in your face as one of thelatest innovations, as I think
many of us are used to, it is purpose built and specialized
and quiet. How can AI tools like Cordy help
(12:25):
to shift that perception? Where there's a lot of alpha
here is a lot about making sure that the last mile is very
valuable. So obviously all these AIS
allows for lower intra barriers and we're seeing some of the
healthcare explosion. We're seeing other markets too.
There's more, I'm sure CRMS thanever before, and many of them
are sure, I'm assuming or being built on AI instead of classic
(12:46):
database structures as they usedto be building today.
That's not different here. It's the same thing in
healthcare. But the real winners here are
getting very, very deep in the work flows.
And I think something that people don't see a lot is
they're actually growing like weeds.
So we have, we're big fans of a Swedish company called Tandem.
There's a Australian company called Heidi Health as well.
These companies are awesome companies and they're growing
(13:07):
like weeds. That's because they really,
really understand their users and they're making all these
decisions to make sure they get them.
And I think that's sort of important point and it's that
allows them to be much more likeAPLG bottom UPS motion.
And I think that's the fear for a lot of people thinking about
doing a healthcare startup. They're thinking, oh, damn, did
I need to do enterprise sales being all these meetings like go
(13:27):
go to a Minnesota and partner upwith some arcade company and all
that stuff. Actually, you can build a
really, really good company. Bottoms up in healthcare today.
PLG works, pickup is real. Doctors are buyers.
They are take forward. They are not the laggers you
think and there are companies around the globe and every
specialty that's growing like weeds right now.
We have a fantastic company we work with Germany called Nelly.
(13:49):
Nelly is one of the coolest company I've seen that's really
owning sort of the dental, dental revenue cycle management
part and they're growing like it's crazy growth as we're
seeing. So I think it's sort of under
the radar a little bit. I think some investors know, but
I think generally people think that still Healthcare is a
smaller live enterprise sales boring, slow.
It doesn't have to be, but it's obviously if if you want to get
(14:09):
the price for it you need to. You're probably stuck here for
more than than average fun cyclein BC universe.
It's so interesting because there is this idea or maybe
misconception that I think is atodds with a lot of tech
companies addressing the healthcare market, where many of
them think more about how, how do I automate Dr. time out?
How do I try to cut down the number of, you know, doctors and
(14:33):
nurses that we need by cutting away these tasks?
And a, that's not going to land super well with these extremely
overworked and very busy medicalprofessionals.
But also, it often misses the mark of what clinicians actually
want to need in order to addressthe needs of their patients and
(14:53):
to address their own workload needs.
So it seems like there are many builders that are
underestimating how little interests many clinicians may
have in AI itself. They, as you point out, just
want software to get out of the way and work for them.
They want to solve a problem like many, many users do.
(15:14):
But often, I think, as we as builders assume, oh, you really
want to get into the details of this?
No, they, we want to solve theirproblem.
They want their problem solved. So how do you bridge this gap,
especially when AI needs to workmore like electricity, seamless,
embedded and invisible? Yes.
So, so our, our job is we're, we're selling AI infrastructure
(15:35):
and models to people who want tobuild healthcare applications
with AI inside, but they want tobe using purpose build stuff
that isn't rate limited, that has high accuracy, that knows
the medical terminology, who's that's billed for embedding into
these systems, that has unique endpoints and SDKS for revenue
cycle management or ambient documentation, all the things
going on in healthcare. And that's where we come into,
(15:56):
we want to be that infrastructure and, and do it at
a price point that's super, super competitive with much less
accurate general purpose models.That's what we hope a lot of
people start building on openingeye or whatever.
And opening eye is a fantastic company.
They'll build fantastic things, But at some point they might
want to get into parts of the workflow, the problems that are
daunting healthcare specific. And there we feel we have a real
opportunity to help. And I think it's, it's let me
(16:20):
down such a rabbit hole as sort of a builder wanting to build
for builders, meeting so many ofour customers customers.
So I had a case not that long ago working with some of working
with this case, one of the coolest, I think fast radiology
companies in the world. So they're doing radiology
equipment and software and we'vemet a lot of their customers.
So actual radiologists, I obviously know what radiology is
(16:41):
before coming there, but I've never like been in, in the room
where they do their work right? And it was amazing work.
And I had heard so many startup pitches in my time, both in the
Oregon and seven in Europe. Like all we're automating away
for like radiology reporting. There's no radiologist anymore.
They don't have a job. I've seen that title so many
places. It's just so funny getting into.
Then you're seeing a person and you've been told by media and
(17:02):
startup pitches like no radiology radiologists wants to
like, examine these images. They just want to get AI to do
it and all that stuff. That's not my experience.
My experience is that we have, in this case, a school of
radiologists who are like Sherlock Holmes.
They can glance at something andthen they go full minority
report on it. They can find small, small,
small Nuggets of gold in there. And they do it like decision
(17:23):
machines all day and they enjoy it.
They're good at it. Could they get AI to help them?
Is there enough radiologists? They could definitely get
somebody to help them. And yeah, there's definitely not
enough radiologists and AI couldpluck some holes.
But the point is, I think the narrative always becomes this
like a plug and replace. I think that sort of the big
missed opportunity both sort of from a capital allocation
problem and from sort of a startup problem, is that if we
(17:46):
really went out there and we're allowed in the room and see how
all this works, the job is a people's job.
It's healthcare. It's not like health task or
health solve. It's healthcare because care is
sort of the continuum you need to do it and a lot of people
working, they actually enjoy it,unlike many people in Silicon
Valley thinks like it's like it's so like non rewarding to be
in healthcare because they've read all the bylines in in in
(18:07):
the COVID coverage. True, COVID was probably not
very rewarding for many because the system was so rigged against
so many providers. But I think a lot of people
still go to healthcare. I have a lot in my family, at
least they went there to have impact, to do care, to make
change, to spend time with patients.
I think an example like this would be all just, again, I know
they're not exactly like the, the, the professionals with the
(18:28):
most patient time, but they're really good at what they do.
So the job isn't here coming in saying you would like you
trained 17 years to be the world's best decision machine,
Sherlock Holmes here, and you'renot doing like you shouldn't be
doing it just be automated away.The job is to make sure that we
can make them be even more productive doing it.
And that might be AI doing some,they do some.
It might be AI augmented, it might be AI doing all the
reporting. That's what we think, but
(18:50):
ultimately it's about listening to where the market is going and
how to deliver more care. It's not to sort of deliver more
continue. And I think in this case, it
actually impacts the, the, the, the capital allocation question.
So the the Silicon Valley narrative as well, because if we
think about it, money will just flow where they get the most
ROI. And obviously great ROI would be
AI does all of it. But in the in get again,
(19:12):
Healthcare is still a care continuum.
So if there's a lot of people doing care, but some of them are
augmented by healthcare, they'remore productive.
That means they have better around a time.
That means more money will flow to these departments that are
more AI enabled, which in turn will mean they can hire more
people than if they can get themas prolific in AI, they will
create more outcomes, more people will flow there.
There'll be more money, more opportunity, better care, better
pay. And all of a sudden the best
(19:34):
people will be working there. And those organizations where AI
enabled will grow. They won't shrink away to like
some weird like dystopian futurewhere there's no care at all.
I love this example of radiologybecause there's such a high
impact specialty and it's clear that adoption is occurring in
this area. Let's unpack this a bit more.
(19:54):
Can you tell us more for our audience?
You may, you may not be aware what exactly is AI doing to
enable radiologist today and andwhere do you see the future?
You mentioned admin work. What else?
So, so I can at least talk aboutthe cases we're heavily involved
in. And I, we are not specialists in
radiology. We have people here who work
(20:14):
here who really know the, the space, but our customers are
world class. And what they come to us saying
is that there's actually a host of products that are sort of
last generation AIS. So classic maybe supervised
machine learning or maybe even just like brute force models
that did stop like dictation. And imagine you're sitting there
looking at the screen at the radio.
It's like say field #2 put that stuff and it's an 8B AT and
(20:36):
there is a this kind of fractureand you're talking medical
language, but you're also talking to an interface.
You're telling denoting where mouse go, where to click, what
to do. Obviously that's not what the
future should look like. You shouldn't be like worried
whether or not it got your commathe right or whether or not it
moved the cursor. Yeah, it should be cognizant
enough to know that you're operating in the system and it
(20:58):
should be agentically enabled enough to go and plug all the
information you're just saying to it in healthcare language as
fast as you wanted, in dialects you wanted, translating into
language like you wanted into all these interfaces and start
all these downstream tasks. That's what we're thinking a lot
about, we call natural language dictation, which is just having
a blast with an LLM copilot thatis AI enabled and agentically
(21:19):
enabled to go inside all these radiology platforms and help
start finish workloads. So you're not just saying move
cursor to point B, you're actually just talking.
And then you have something thatfeels cognizant that isn't sort
of an after the facts, like few shot, one shot LLM writing a
summary that's riddled with mistakes.
It's something that happens in real time.
That is sort of a recursive reasoning and making sure that
(21:41):
it actually finds the facts, understands the facts and starts
the actions needed. What's a more productive
narrative to leverage here as AIcompanies think about adoption
then within emerging healthcare verticals?
Obviously understanding your vertical, really getting deep in
(22:02):
the language of it and not trying to sort of, I've heard so
many health healthcare professionals talk about sort of
the, the health like the tech savior complex.
It's like, don't land with the savior complex of you solving
anything. Like we all humbly need to know
that these are like, these are people who usually chose their
career path because they're among the smartest of us and
we're just there to serve them and make that happen.
(22:23):
So I think the narrative is, wait, wait here.
It's just to make sure we get away from the dance floor and
make sure, like, the food is served and the light is on and
everything works. And the people are really good
at those workflows. They are the people we see
growing like weeds. Enablement, not replacement.
Absolutely. This reminds me of the
conversation around AI and different creatives where yes, I
(22:46):
in that instance, there is I think a side narrative of, oh,
we're going to enable everyone to, you know, do more creatively
by giving them AI tools. But often expert artists don't
really want AI to paint for them.
They enjoy the act of creation. They maybe want to be enabled in
areas that are around their key tool sets.
(23:07):
Is this a fair parallel do you think?
I think it has a lot of overlaps, that's for sure.
And I think the respect for for the work is key.
And I think that's very much sort of built into some of the
critique. And so if it's some of the the
sort of the phenomenal headlinesof creative sort of going away,
I think Mira Muradi at some point said when she was at Open
the Eye that they will replace alot of creative work.
(23:29):
And some of those creative jobs might never, should never have
been there. I'm sorry, Mira, if that's not
correctly quoted, but I think the point was just obviously
there's a building critique herethat like if it can be replaced,
maybe it should have been in theget go.
I just do think, especially for creatives, like I have never
seen a chess match between 2A ISI have no plans of seeing like
a Dota match between two Dota playing AISI have known, but a
(23:51):
Dota like world class team of five playing a tech team of five
AISI might TuneIn that might be fun.
So I think this is a lot about the interplay.
It's not about the pureplay. AI is going to be cool and a
Genting AI in Healthcare is going to do so much work that
nobody signed up to do. I promise you, very few people
in the world find like billing, coding, fantastic endeavor.
(24:13):
Some do and they're amazing and we need them.
We'll always need some, but they're rare.
They're rare and they should be rare because it's hard work and
it's complicated. And maybe those people who are
doing it today who don't love doing it, they could be doing
patient facing stuff instead. But that's I think the
opportunity of healthcare. Yes, we'll replace some work,
but for the majority of it is way more about open ending or
enabling. I like this idea of
understanding the intrinsic motivations of professionals and
(24:37):
people in order to tailor what we're building to them.
And I think it speaks to being customer obsessed and being user
obsessed in a way that too oftenfolks who come in with a savior
complex, as you put it, or maybearen't listening to the people
on the ground, but instead are saying, oh, here's how I can re
(24:59):
envision this market. And while it's great to have a
vision for the market and there's significant
opportunities to enable physicians in particular, 24% of
whom think about leaving their role weekly due to overwhelming
workloads. You know, clearly there's a huge
opportunity there. But failing to talk to them and
understand their pain points is just a huge mistake.
(25:22):
And it is very clear to me that you're not doing that.
And you're both addressing the needs of your, your own clients,
the builders who are enabling healthcare providers and then
also digging into the individuals within the system.
Yet there is this significant workforce challenge.
You know, we, we mentioned there's challenges of the admin
work, there's challenges with what I've heard you refer to as
(25:45):
pajama time and the lack thereofof, of folks doing notes at
night. And there's a gap in potentially
10 million trained healthcare professionals that will need by
20-30 that likely can't be filled through traditional means
alone. So what needs to fundamentally
change in our approach to AI development and deployment for
it to truly address both the looming healthcare professional
(26:10):
shortage? I mean, frankly, there is one
already and also the pains that individual providers and
physicians are feeling as they navigate this challenging high
stakes regulated environment that's extremely high stress.
It's such a good point and I think The Who actually upgraded
their number of the the caveat or the delta of of the
(26:32):
healthcare professional will need by 20-30.
It's hovering between 10 million.
We lack just to deliver the carewe delivered today, not the the
cooler version of healthcare we'll have learned to do by
then, but just to do what we do today.
It's at least 10 million we'd like globally healthcare
professionals. So you're right, there's just a
problem if you can't do what we do today.
But if we look backwards and we assume it's the same curve of
innovation and novelty, we probably don't want to do the
(26:54):
healthcare we do today in 2030. We probably want to do way more
preventative care, longevity, all that stuff, right?
So you're right, there's a problem and I think sort of one
of the things we're seeing sort of what's the sort of the low
hanging fruits like in energy weall want like maybe some of us
wonder fusion or some of us wantfission, some of what want
something like solar. But some things are like easier
to do like solar and something are harder like fusion.
(27:16):
And I think in our case in healthcare, if we just look at
the 25% of the the workflow, so I said before that we have sort
of 40% issue of work done is like language based.
If we look at sort of workflows that are language based in
healthcare that are ripe for AI innovation and we just take the
25% most ripe and we add AI to the parts of it that could be
automated. I think it's McKenzie or
(27:38):
etcetera at McKenzie. That said, if we do that, we can
actually move $1 trillion worth of HCP salaries away from admin
that's in workflow like collecting data, process data.
That's like the, the, the catch all kind of workflows, right?
If we did AI optimize those, we have up to a trillion dollars.
(27:58):
We can move back into care or somewhere else, but back into
care seems nice, especially if we lack 10 million people.
And if you look at sort of the what does like 1 trillion,
obviously it's very different salary curve in, in parts of
Europe versus US and so on. But if we sort of look at the
the the gap here is actually possible to map quite a bit of
the most critical ones by just doing the most optimal, most
(28:21):
obvious reallocation of time by using AI healthcare.
Let's drill down on solving thisproblem because you've talked
earlier in our conversation about this major opportunity
around, you know, 4 out of 10 hours in healthcare are spent on
language tasks. So this seems directly
applicable to AI as it stands today.
(28:44):
And then, I mean, the scope of this is staggering, obviously,
you know, potentially moving a trillion hours back in the
workforce or, or freeing up those hours.
How would you break down how this would all work?
We need a Cambrian explosion. We need so many more
opportunities for healthcare professionals to leverage these
(29:08):
kind of technologies. Today, we have a lot of
healthcare still running on premise.
I think that's sadly the truth. If we look at the big, big, big
EHR platforms like Epic, a lot of it is still on premise.
We're seeing it move more and more to the cloud.
And I think Microsoft and Amazonand many other cool companies
are playing a large role in making that happen.
So we will unlock more and more opportunity to add more of these
types of technology as healthcare moves to cloud.
(29:31):
Secondly, we need to make sure that we find better ways of
enticing the big sort of systemsof record to start opening up
their doors. Some companies like Epic are
like they're doing all sorts of initiatives now to open up the
doors to more AI vendors. They're writing AP is they're
opening up. And I think that's the path
forward. Some of our customers like
(29:52):
Dataloose, the world's first biggest DHR platform, they cover
patients from Bolivia to Germany.
Like it's a massive undertaking.And they have a massive tech
team that are doing a lot of this stuff to enable more and
more tools to plug in. So more cloud, more plug in,
more tools, more builders that dare specialize.
That in turn I think will be theCambrian exclusion.
And if we can then get some AI that doesn't hallucinate as
(30:13):
much. I think the healthcare
professionals actually want to adopt it, even in fax riddled
Germany. How are you addressing the
hallucination problem? What's your approach to
evaluations? I wish that was the only thing
that relaxed to make sure we we we have opportunity to
hallucinate less. The the boring answer is a lot
of tedious work across the factory 4.
(30:33):
So if we think of ourselves as afactory 4, we input health data,
we output health data or health reason.
No. And, and in that factory 4, we
have many models, many machines.And at all times we need to 1st
off, we need to benchmark. Was it the quality of all the
data we got in? Did we understand it?
Did it, did it translate well? If we transcribe it, did we
transcribe it? Well in the factory, every
(30:54):
component that's interacting hasto have benchmarks.
So we need to understand, removedata.
What happens? Where does it happen?
We also need to understand what's the contribution of this
machine in the factory 4, so we could plug it and play it.
Oh, somebody built some really cool shit over here.
We're really inspired by this company.
How do we learn from finally, when we output it, we always
have public benchmarks and public benchmarks to be honest
(31:14):
is for the vast majority of not very useful.
If you go to the Hugging Face and find some of the big sort of
medical benchmarks, pub Med, there's many others.
You look at the the leader, the scoreboards of many of these
health data sets. The leaders are non commercial
models. I think everybody in machine
learning will have a good guess at what that means.
It's probably because it's not very useful in practice and
(31:36):
there's probably really good reasons to not bend and hammer
on anyone. But there's probably really good
reasons why it was trained in a way that makes it really good at
tests, but not very good in the wild.
So I think it's not just benchmarks.
It's an entire infrastructure, it's many pipelines.
It's understanding like not all data is created equal, not all
data processing is created equal, and every component in
managing this ecosystem of reasoning is a massive
(31:59):
undertaking because we need to be able to actually explain what
we do, why we did it, why we changed it when we did what's in
the model cars, like what's in the food we're serving, what's
the ingredients we used, why didwe use them, when did we move
them? How do we launch it, where is it
hosted, at what level and what state are we encrypting?
There is so many parts of this puzzle that it's to to many
(32:20):
dismay if they want to play in the infrastructure, land,
healthcare, it's sadly not just one thing and there's not just
one benchmark. It is an entire infrastructure
of solutions that has to come together.
I think that's a big learning from RCA is like you can win
healthcare by being good at one thing.
You have to be good at many things at least quite a bit, and
then you can be OK at some. Then you can decide to not be
good at all at some others. Like us.
(32:40):
We're not very good at the last morale, the workflows.
We have partners who are fantastic at that part, but we
want to be really good at infrastructure part.
I think it's wonderful that you've found the focus of where
you want to approach things. And I appreciate the the detail
that you're bringing to both explain ability and
(33:02):
observability as well as customization.
Because too often I think we areseeing approaches in different
verticals with AI where you justsay, I'll throw a big
generalized model at it. It's going to work great.
And in fact, you need to really customize the metrics you're
(33:22):
using, the benchmarks you're applying, how you approach the
task itself. You know, what tools are you
calling within an agentic system?
Everything needs deep customization and infrastructure
work to effectively address problems in in challenging
industries, particularly for highly regulated ones.
You know, when you introduce nondeterminism, there is an added
(33:44):
element of of risk, which an added opportunity to but there's
there's so much to solve there. So you know, I'll say a Galileo.
Obviously we are a huge fans of this approach for customizing
benchmarks and metrics and really building hybridized
systems that include both that like expert feedback, but also,
(34:04):
you know, leveraging LMS or SLMSlike our Luna models of judges.
So I so I love that Cory's taking the same thought process
of, OK, how do we make this explainable for our customers
and for our users? Because without that, it's hard
to really deliver value, particularly in these very
consequential workflows. And then how do we customize
(34:26):
this to really meet their needs instead of just saying, Oh yes,
here's this generalized benchmark that we can apply.
So love that you're thinking that way.
I think that's absolutely the right approach, but I also
imagine there are major hurdles that you're facing, whether
they're technical, regulatory, cultural, to achieving the level
of trusted automation in healthcare that I hear you
(34:46):
seeking to achieve. How does Cory take a compliance
first approach to help address these?
Do you know what's really helpful?
Being from Europe, we're really good at compliance.
We're really good at it. So if I were doing AI for for
many other things in the world, I wouldn't be saying this.
It would be an inhibitor. But we actually have a really
(35:08):
cool, very, very, very large US customers who are launching very
soon. I don't exactly know when this
is going live, so I won't name them yet.
They came to us. They had a very, very long
purchasing site. They're a massive company.
And one of the things they like was like, OK, so you guys are on
like AI Act and GPR and all thisstuff.
And they're actually moving towards buying from AI vendors
who are more in healthcare, in healthcare.
This is healthcare aligned towards stuff like AI Act
(35:30):
because they think no matter what the market will drift
towards some of this stuff because yeah, some of it is like
red tape, we should remove it. But there's also a nucleus of
truth back to you want to be able to trust it even in the
high stress environment. And I think yes, there is a lot
of red tape in Europe. That's not great, we all know,
but there's a lot of red tape that really makes sense and a
(35:50):
lot of furrow People really thought about it.
And I think that's something that coming from Europe, having
built a company grounds up here,like our first many, many models
we like we pre trained, we did all of ourselves.
So we really, really know how tomake a good soup.
And I think that every chef's knows if you can make a good
soup of simple ingredients, you're on to something.
But also allows us to really control went into the models.
How do we train them? How do we like structure it?
(36:11):
What do we do post training? If you actually do the work,
it's much easier to give somebody else a recipe.
I definitely agree with you about the ability when you have
already addressed certain regulations and compliance needs
to then apply that approach. Being able to have that baked in
from the ground up is so important.
And I'll say on our end, we've seen this in financial services
(36:33):
and banking already where, you know, because we're already
working with JP Morgan Chase, because we're already working
with Citi, it becomes so much easier for us to have these
other conversations with major banking and financial service
providers because we get their regulatory concerns.
We know the customization needs they have.
And I'm sure that's the case foryou across Europe, across the
(36:55):
world, as you are able to apply this regulatory framework that
you've you've already understoodand become compliant with.
I'm going to give you a leg up in consequential industries.
I think we see this also with, you know, AI defence and
elsewhere, but how do you get AIto a point where it genuinely is
(37:15):
reducing workload? Like, great, we're, we're
complying with regulations, we're, you know, handling the
different situations we need to,but how do we make sure it's
invisible and reliable? There's so much infrastructure
work that goes in behind that. What are you working on at Cordy
to support this mission? Thanks for asking this.
(37:36):
This is where you can keep me all night.
An example is in healthcare right now, everybody's pretty
hyped about AI Scribe. So these like ambient tools,
you're in sort of outpatient care.
You go to your doctor and they turn on a mic or their phone and
they start recording. And then there's another them
listening, recording. And then when it's done, it sort
of spits it into this data pipeline.
It transcribes, then song rises,then you blurb out some text and
(37:58):
here you go. Your note is done.
And always today these like notetakers are like they're saving
so much time. Yes.
But but we, we did a survey of 2000 of the earliest adopters of
these cool tools in the US and they told us on average, they
spend at least three hours a week just like betting and
correcting their AI. And they say it's a constant
worry for them whether or not the AI had like set, shared or
(38:19):
done something they like had miscollect, mis recollected or,
or didn't change. So we're creating workflows here
that shouldn't be there, right? That's not the promise of AI is
that like we'll give these fantastic HTPS these new tools
and then they'll sit and get worried about those tools like
they're burning out. We should not get them more
worried, right. So yeah, they're great.
We fought like that's exactly the kind of thing where like an
(38:39):
infrastructure company like us built for it can make a change.
So so we we offered up somethingcalled facts are.
So it's a it's a recursive reasoning pipeline.
So it's not just sort of an LLM that one shots or few set
documentation, but it actually instead listens in.
So you turn on the microphone. You've maybe before you build in
some like general purpose LLM, alot of model whatever.
Now you have Corey instead of just waiting and like batch
(39:02):
processing a few shutting it, we'll listen in all the time.
So you open it. And while you're talking to the
patient, we are doing recursively revisiting with
using like an orchestra of of anelims.
What is the sort of saline factsthat are really important and
can we revisit their importance as the conversation changes?
We all know, like we've been having a conversation like this
(39:22):
or we're having a conversation over wine and subjects change.
But in a transactional conversation, like in
healthcare, you're transacting where to go, where to stay on
medication, to do. There is a kernel of truth that
needs to be respected. And all of the side time we
visited. But if you just do batch
processing, these little limbs will spit out shit.
They will anchor on stuff. They will have like indexed on
something. It will maybe even hallucinate.
(39:45):
We're all the time trying to make sure that there's an
orchestra and there is a series of models as judges passing
judgment on whether or not this reasoning pipeline gets it.
And what we're seeing when we benchmark it on on some of the
big public benchmarks is that not only will we find 49% more
of what matters, the really important healthcare
information, we will also reducelike the verbi verbose sort of
(40:07):
classic. We all know in the limbs can be
pretty verbose, right? So we can reduce 88% of what
doctors would deem verbose or noise in these documentation.
And if you actually try to test whether or not they like the
doctor, both find it like accurate, concise and complete,
complete being. This was something I would have
written. It's like up to my standards.
When they work together with a real time flow, like recursive
(40:28):
reasoning from our effects, our model they actually feel is as
complete or close to as what they did, which is a vast
difference when there's a limb spat out something that didn't
feel theirs. So this is the kind of thing
where we can make real decisionson the infra layer that makes
the like the the the entire experience for our customers.
Customers completely different without it, like at the end user
(40:49):
stage feeling much different. Let's dig into this infra layer
a little more. I'm curious to understand more
about how you're applying judgesto double check outcomes and
evaluate what. What type of judges are you
using? You mentioned using multiple
judges. How are you waiting them?
So some of it that we actually have in our documentation right
now, which is live on, on on ourwebsite, you can go build it.
(41:11):
And I think some of our customers might have an impact
on how we want to do it in some of our enterprise deals as well.
Obviously, the challenge is to find a scalable way of
understanding how to pass judgment on something that is a
moving target, which is hard. So we did, we do need a lot of
data sets. We need a lot of holdouts, which
luckily we've been here for almost a decade.
So we have done a lot of that. And then it's about really
understanding the workflow it's embedded into to make sure we
(41:33):
applied in the right point at the right place.
Furthermore, at the end of it, we're lucky in our workflow that
a user has to pass judgment, right?
So a doctor has to say what you did was cool.
So we actually, at the end get aquite cool golden label of
whether or not it's as complete or as concise as somebody would
actually use. It regular human feedback to use
for reinforcement learner exactly and.
And what we even see is that when you normally get these like
(41:56):
few shots and you sit in the endwith this like build wall of
text, like the, the quality and the feedback is obviously
temporarily it's all batched, right?
Whereas if you're all the time, you can just like, instead of
sitting there at night pyjama time writing and rewriting it as
you recall it, we just like glance your screen and you can
like click, click. Then you've given it feedback
(42:16):
that actually adds A temporal element as well that we can
stand and use. And all of the sudden we can
stream the entire conversation for the API, understand as well
the interaction with the AI as it moves.
And all of the sudden we don't just have a really good golden
label of concise and completeness.
We also have an understanding oflike when you interacted with
the API for your platform, what happened?
What did you do it? What did you interact with?
(42:37):
What did you change? So all these things helps a ton
looking. Ahead, as AI models become more
capable, where do you see the next significant opportunities
to move beyond administrative tasks and begin to safely and
effectively support diagnostics,decision making?
And ultimately not only drive improvement and freeing up time
(43:02):
of the clinicians, but also directly improving patient
outcomes in a more, you know, regulated and compliant way
while still respecting the clinician's role.
I think it's a lot about sort ofif we took by to it, there is
something that's like new workflow, new market.
And in that like new workflow, new market, there is stuff today
(43:23):
that hasn't happened in healthcare yet because we might
not afford it, It might not be obvious that should be happening
in the future. And I think there's some cool
companies out there trying to map that space right now.
Some of it is like like existingworkflow, existing market that
might be like adding AI to automate a ways sort of
secretaries. I think a much more interesting
workflow is like how many conversations could you have?
Or me for that matter. Let's say I picked up a new
(43:44):
medication. I have never taken it before.
Hopefully I'll never do it again, but that means I have 0
contextual there. My physicians usually don't have
time to call me and select. Tell me more about that.
Like that like weird feeling of like airiness you get in your
head. Like it's a soft signal.
It's not probably not a leading indicator.
It will go away so nobody will call you, but if you're the
(44:04):
average cost per minute of high quality medical reasoning with
solutions and infrastructure like us can be like orders of
magnitude cheaper. Why wouldn't you push amazing a
is at calling all your patients?If that's in pharm eyes, that's
in like physical rehab. It's like psychology,
psychiatry, primary care, outpatient care, you name it
home care. Why would you spend way more
(44:26):
time just reasoning with the patient, cautioning them,
helping them, guiding them if you could and you actually had
an infrastructure that was safe enough to not to lose any
weight, take all the pills, it'sgoing to be great, right?
We all heard the like Google like offering patients to eat
one rock a day because that's like a fictive study from
Berkeley has said that's great, right?
If we can control it, then like all of the sudden bring compute
(44:49):
costs down will be such an unlock for like you work for the
new markets that we haven't thought about yet.
Andreas, thank you so much for staying up late in Copenhagen
today and joining on the podcast.
It's been such a pleasure havingyou on the show.
Where can our listeners go to find out more about Cordy and to
learn more about your work? Hey, thanks and thanks for
having me. I really enjoyed the the the
conversation. We're at Cordy dot AI.
(45:12):
Go sign up, ping me on Twitter or LinkedIn.
My handle is Andreas Cleave or ping us at Cordy.
Do it on Instagram. We'll, we'll shoot you over some
credits. You can start building if you
have a cool use case where you need something bespoke,
something hard. If you're a University Hospital
wanting to do real clinical research with an AI partner.
We're, we're not sort of the latest fad.
(45:33):
We're here to stay and we'd loveto build cool shit with you.
I love it and I honestly have super enjoyed this conversation.
It's been such an interesting look into this.
I think misunderstood vertical as you put it earlier in our
conversation. There are 1000 different
verticals within healthcare and they all deserve their own
(45:54):
treatment. There's such an opportunity
here, and it's inspiring to hearthe stories of companies like
Cordy that are focusing on highly regulated consequential
verticals and making deep inroads by being considered
about their approach. So tons of great lessons that I
think our listeners can take away from the approach of
QWERTY. And hopefully a few folks on the
(46:18):
healthcare side of things will be reaching out to you after
this. So we'll certainly link
everything that we've discussed today in the show notes.
Super excited to learn more about your evaluations and how
you continue to build out the QWERTY platform.
There's clearly such an opportunity here.
And for our listeners, if you enjoyed this episode, support
the show by subscribing. Give us a like a comment, a
review. It matters a ton.
(46:41):
Take a little pajama time to take two or three minutes and
just, you know, drop that comment.
Those drive the algorithm in ourfavor, whether it's on LinkedIn,
whether it's on Spotify, YouTube, wherever else.
We deeply appreciate your support.
Thanks again for tuning in. And Andreas, thank you so much
for joining us. It's been a pleasure.
Hey, it's been a pleasure and good good work on all fronts.
Big fan of yours. Thank you very much.
(47:01):
Appreciate it.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
The Breakfast Club
The World's Most Dangerous Morning Show, The Breakfast Club, With DJ Envy, Jess Hilarious, And Charlamagne Tha God!