Latest / Becker’s Payer Issues Podcast / Breaking Free from Legacy: Modernizing Health Plans with AI and BPaaS

Transcript

Breaking Free from Legacy: Modernizing Health Plans with AI and BPaaS · Becker’s Payer Issues Podcast

0:00
Hi, everyone. This is Lucas Voss with Becker's
0:02
Healthcare. Thanks so much for tuning in to
0:04
the Becker's Healthcare podcast series. It's fantastic to
0:07
have you. An exciting topic today, very wide
0:10
ranging, the challenges of costs in the payer
0:13
market, and nobody better to talk about it.
0:15
So excited to have him on for today's
0:17
discussion. Kevin Adams. He's the CEO of HealthEdge.
0:21
Kevin, so great to have you. Thanks for
0:22
being here. Pleasure to be here. Looking forward
0:24
to the conversation.
0:25
Absolutely. I do wanna start off for our
0:27
audience with just a little bit of an
0:28
introduction.
0:30
If you could just talk a little bit
0:31
about yourself too and but also about HealthEdge
0:33
and and what you do on a regular
0:35
basis in health care.
0:37
Absolutely.
0:38
So Kevin Adams,
0:40
I'm relatively new to the role of CEO
0:42
of of HealthEdge.
0:44
It's been about sixty days since, Bain Capital
0:47
acquired,
0:48
the prior company that I I founded,
0:51
and combined us with HealthEdge.
0:54
I've been, in the health insurance space specifically
0:57
as exciting as it is for the last,
1:00
twenty six or twenty seven years,
1:03
came out of Wall Street,
1:04
into what I thought was a
1:07
area of
1:09
of need
1:10
Mhmm. And have spent the last
1:12
twenty seven years trying to help,
1:15
Americans across the country gain better access to
1:17
health insurance.
1:19
The last ten years,
1:21
we we,
1:22
ran a company called USD health proof.
1:26
That company was focused on outcome based services
1:30
and technology
1:31
for health insurance company
1:33
companies where we were running the back office
1:36
and,
1:37
deploying
1:38
next generation technologies to support modern day products,
1:43
networks, benefits
1:44
in those health insurance companies
1:46
had about,
1:48
25
1:50
customers where we were running our BPAD solution.
1:54
And, I'm sure we'll, we'll talk more about
1:56
that later, what that is. And
1:59
one of the key partners
2:01
in our efforts over the last ten years
2:03
was a company called HealthEdge.
2:05
And so HealthEdge provided,
2:07
much of the
2:09
software
2:10
that we were leveraging for our solutions.
2:13
And,
2:15
we became, obviously, a critical partner to HealthEdge
2:18
for deploying their software
2:20
and,
2:21
just so happened to have been their largest
2:23
customer.
2:24
And so,
2:26
in
2:27
in early twenty twenty five,
2:30
we went to market along with health edge.
2:33
I'd like to say that was well planned,
2:35
but luck sometimes is is a great benefit
2:38
also.
2:39
And,
2:40
so health edge and health proof went to
2:43
market in early twenty twenty five,
2:46
and,
2:47
we were able to work with a number
2:50
of private equity firms. But at the end
2:52
of the day, Bain Capital was not not
2:55
just
2:56
the best
2:57
solution
2:58
for from a standpoint of having understanding and
3:01
direction and desire,
3:03
but,
3:04
also happened to have been the the best
3:07
price for both health edge and health proof.
3:09
And so at the end of the day,
3:11
they won the the process.
3:13
And for both Healthproof and HealthEdge and and
3:15
effective,
3:17
September 6,
3:19
put the two organizations
3:20
together, and,
3:22
I was either
3:23
blessed or cursed to become the CEO of
3:25
combined company. So I'm sure it's it's blessed
3:28
in this case. Yeah. Yeah. Exactly. It's a
3:31
awesome awesome honor,
3:33
and and we have some absolutely amazing owners
3:36
in Bain Capital.
3:38
And, we are truly looking forward to the
3:40
future that we can we can bring to
3:42
The US health insurance market.
3:44
I do wanna
3:45
level set us a little bit in terms
3:47
of where we're at today from a market
3:49
perspective, but also in general from your perspective
3:51
at HealthEdge.
3:53
What are some of those biggest market operational
3:56
and and tech certainly technological
3:58
challenges that
4:00
US health plans are facing today?
4:03
Wow. I'd say, a wide range of challenges.
4:07
And I would say number one,
4:10
you know, nontechnical,
4:12
just the reality of health insurance,
4:14
is that you are cost wise impacted
4:18
on many, many different fronts. I think you
4:20
see every day, MLR utilization
4:23
is through the roof for for The US
4:25
population,
4:26
whether that is a,
4:28
effect
4:29
from COVID or not, meaning,
4:33
the results of people not going to the
4:35
doctor for a long time are now manifesting
4:37
itself in terms of higher rates of disease
4:40
or higher rates of utilization.
4:42
And,
4:43
that is having an outsized impact on on
4:46
health insurance companies bottom lines.
4:50
Not a lot you can do whether it
4:53
be technology
4:54
or otherwise.
4:55
That's just a reality.
4:58
The second thing in the l m l
5:00
r portion of the the cost
5:03
is
5:04
that providers,
5:05
oddly,
5:06
because their systems
5:08
are
5:08
somewhat easier to
5:11
to manipulate and use,
5:14
are able to leverage AI
5:16
slightly better. Actually, I would say measurably better
5:18
than the health plans have been able to
5:20
leverage AI
5:21
and are
5:23
increasing the accuracy of coding
5:26
through the leverage of AI. And that increase
5:29
in coding is causing a higher
5:32
medical cost also for health plans. It's about
5:34
a you know, I've heard numbers vary from
5:37
20 to 30% of the increase in MLR
5:40
is coming from coding accuracy.
5:42
A lot of people
5:43
tend to say
5:45
upcoding,
5:46
but I I think that's a it usually
5:49
held as a negative statement, meaning upcoding is
5:52
trying to,
5:54
maliciously
5:55
add code that shouldn't be there. This is
5:58
just accurate coding.
6:00
And so where the providers were losing revenue
6:03
because they were unable to accurately code their
6:05
claims,
6:06
now they're able to increase the accuracy through
6:09
AI, and that's driving 20 to 30% increase
6:12
in MLR costs. So you put those two
6:14
things together, MLR
6:16
is a huge impact to to health plans.
6:20
And the reason I say that is because
6:22
the MLR portion is a very, very difficult
6:25
thing to impact in the short term.
6:28
And so when you're looking at the health
6:30
plan, looking for ways to lower your annual
6:33
or year to year cost, the administrative
6:35
side of your system is about the only
6:38
thing where you have immediate or very near
6:40
term opportunity to reduce cost.
6:43
And the administrative side is where everything around
6:46
claims processing, provider data management, enrollment, claims intake,
6:50
all those things, provider payments,
6:52
all those things are where you need to
6:54
find a way to maximize
6:57
your efficiency
6:58
and effectiveness
6:59
because that's an immediate opportunity to reduce cost.
7:02
And so that's kind of laying out the
7:04
groundwork
7:05
of of where the health plans are today
7:08
and why,
7:09
this business happens to be something of interest.
7:12
And then I'll kinda step back as to
7:14
why the administrative
7:16
side of health plans are inefficient and ineffective
7:19
and ripe for for opportunity to help them
7:22
reduce costs.
7:24
And so, you know, if you look back
7:27
over the last, you know, I'll tell you
7:28
tell you, out of my career, the last
7:30
thirty years,
7:31
you know, it was really about thirty years
7:34
ago when the technology
7:37
used to automate
7:39
manual claims processing
7:41
began to come to the forefront.
7:43
And I'm not saying it didn't exist before,
7:45
but it was really starting to become,
7:48
you know, computerized.
7:50
I'll use that term,
7:52
computerized
7:53
in around 2000.
7:55
And a lot of those systems that were
7:57
implemented were predominantly
7:59
mainframe based systems,
8:01
large mainframes,
8:02
or in in the case of the regional
8:04
plans, they might have been on mid tier,
8:06
you know, a s four hundreds and that
8:08
kind of stuff. Still mainframe centric, COBOL based,
8:12
and monolithic in nature. And what I mean
8:14
by monolithic in nature is that everything around
8:17
the back office of process health insurance
8:19
was on a single piece of
8:22
of software. The Yeah. The in quote claim
8:25
engine was So think about that for a
8:27
second. I think that that's a very, very
8:28
important piece to think about. Right? One single
8:30
piece of software. And also the time frame
8:33
that you've mentioned, this has been on go
8:35
twenty five years. Correct. Correct.
8:37
You know, sometimes I look at this and
8:39
say, you know, I've had a great career
8:41
and I've learned a lot, but clearly I'm
8:43
a failure because it's taken twenty five years
8:45
to get to the point where I think
8:47
we could actually
8:48
actually execute against the the the dream that
8:51
we've always
8:52
had. But,
8:53
anyways, that monolithic piece of software in many
8:56
large health plans still exist today
9:00
and, is a very costly, very complex
9:04
capability
9:05
to replace.
9:06
And it's basically when you try to replace
9:08
the claim engine in a health plan, you're
9:10
doing open heart surgery on that health plan.
9:14
Mhmm. And not only you're doing open heart
9:16
surgery, but it's not like you can stop
9:18
the blood flow.
9:19
You have to actually keep the blood flowing
9:22
while you're replacing this major organ.
9:25
And, you know, any any,
9:28
challenges, mistakes,
9:29
or issues
9:31
can result in the death of the health
9:34
plan.
9:35
Certainly, the end of a career for CEOs
9:38
and CIOs and whatnot. So
9:40
it's a very critical, very difficult thing to
9:42
do and a lot of resistance,
9:45
hence why there's still many,
9:47
many opportunities out there,
9:48
is that it's there's never been a company
9:50
that could effectively,
9:53
execute against a strategy to replace this and
9:56
have a high degree of success in doing
9:59
it measured over the last ten years. And
10:01
that's what USA Health Proof was that we've
10:03
been able to execute
10:05
in in that time period 25 different health
10:07
plans where we've been able to effectively move
10:10
them from legacy to modern technology.
10:13
So the the just kind of cycling back
10:15
to the problem is that that monolithic technology
10:19
could not sustain
10:21
the needed improvements in
10:24
network flexibility
10:25
or reimbursement
10:26
changes, value based reimbursement,
10:29
benefit you know, incentivized benefits or complicated benefits.
10:32
And so people over time
10:34
started to
10:36
plug in,
10:38
what they would call as modern technologies
10:41
around this legacy footprint,
10:43
and you ended up with the micro segmentation
10:47
of the business processes across
10:49
multiple different technologies
10:52
that were all wired through bubblegum and bailing
10:56
wire into this legacy platform.
10:59
And it added to the complexity
11:02
while giving a little bit of measurable benefit
11:04
in terms of capability.
11:06
At the end of the day, you ended
11:08
up through the segmentation of the business process
11:10
by having, you know, a specialist business unit
11:13
that's running enrollment, a second business running, you
11:16
know, pricing, a third one running,
11:20
claims reimbursement.
11:22
So all of this the segmentation
11:24
caused complexity in the environment,
11:27
and
11:27
that complexity
11:29
led to additional resources needed, which at the
11:32
end of the day ran to higher cost
11:34
to operate your back office.
11:36
And so naturally, the question is is how
11:39
do I get rid of this complexity?
11:42
And people still
11:44
think back to,
11:45
well, in the old days, I would buy
11:47
a monolithic claim engine and boom, you know,
11:50
I've solved all my problems. Well, that doesn't
11:52
exist anymore.
11:54
When you look at a claim engine today,
11:56
a la HealthEdge, HealthRoots payer,
12:00
and some you know, the equivalent is the
12:01
same across all the competitors is that they
12:04
are only
12:05
a small portion
12:07
of the overall
12:09
technology set required to run the back office
12:12
of a health plan, and it requires an
12:15
ecosystem
12:16
of technologies
12:17
and services
12:18
in order to do that replacement
12:21
modernization
12:22
of the back office of a health plan,
12:24
which ultimately is what drives the substantial cost
12:27
savings that health plans need. Yes. And you've
12:30
you've mentioned the the the sort of the
12:32
solution to this in terms of the the
12:35
modernization part. And I I I'd love for
12:37
you to you mentioned an acronym in the
12:39
introduction too that I wanna get to because
12:41
it's very important to understand the this modernization
12:43
piece, which is business process as a service.
12:46
Right? Yeah. Yeah. And also business process outsourcing.
12:49
There's a key difference. There is. And I'm
12:51
wondering if if you could get to where
12:53
we are today in terms of modernizing.
12:55
First of all, explaining what those two are,
12:57
how they're different, but also how they're addressing
13:00
those challenges in turn. Yeah. Absolutely. So, you
13:03
know, I think the you know, 90 plus
13:06
percent of health insurance companies
13:09
leverage business process outsourcing.
13:12
And business process outsourcing, the way I would
13:15
term it is mess for less.
13:17
So you are taking the existing technology,
13:21
the existing processes,
13:23
and you're leveraging labor arbitrage
13:26
to do that same work at a cheaper
13:29
price.
13:30
You cannot
13:31
inherently reduce
13:33
the operational efficiencies
13:36
of
13:36
what is going on because it's the same
13:38
method you had before.
13:40
The only thing you can do is reduce
13:42
your arbitrage. You know, you can arbitrage your
13:44
costs, I e, outsource to offshore because that's,
13:47
you know,
13:48
the only way you're going to achieve that.
13:50
And so set aside,
13:52
you know, some of that negative connotism. I
13:55
mean, it it is what it is. That's
13:56
a reality across America. So I'm I'm not
13:58
gonna, argue for or against it.
14:01
But you can only get
14:03
somewhere between 1015%
14:05
savings.
14:06
And because it's still ultimately mess for less,
14:09
the outcomes
14:11
that the vendor,
14:13
is going to give you
14:15
is never gonna be aligned with the outcomes
14:17
that you need, which is long term
14:20
continuous reduction in cost.
14:22
Because if the vendor
14:24
is executing against that long term need, then
14:27
they're reducing their revenues.
14:29
And no no
14:31
no vendor is interested in reducing their revenues.
14:33
They're always trying to increase the revenue or
14:36
or increase the profit.
14:38
And and so,
14:39
you know, BPO
14:41
to me was never a long term solution.
14:44
And so what we came up with was
14:47
that you need to improve the underlying
14:50
business
14:51
technology
14:52
and business processes
14:54
that run on that technology
14:56
because that's the only way you can drive
14:59
ultimate real efficiency.
15:01
And then you can leverage arbitrage on top
15:04
of that to double down on the effectiveness
15:07
of of this new technology.
15:09
And Replacing the foundation sort of you're building
15:11
a house. Right? You're you have to instead
15:13
of fixing everything that's inside without fixing the
15:15
foundation, you have to fix the foundation first.
15:18
That's a that's a great way to look
15:19
at it. A great way to look look
15:21
at it. And and so when we,
15:23
in
15:24
2016,
15:26
started our business at at Healthproof,
15:28
we understood
15:30
that, number one,
15:32
the challenges
15:33
of replacing
15:35
a core administrative
15:36
back office were immense.
15:39
And there had to be a way to
15:41
simplify that. Otherwise, nothing we were gonna do
15:44
was ever gonna be effective or as effective
15:47
as it it should have been. And so
15:49
we created a methodology
15:52
and a solution
15:54
that was an ecosystem of technologies
15:57
prewired together
15:58
so that when we deployed
16:00
the solution, it wasn't let me deploy 15
16:04
different pieces of a solution and figure out
16:06
how to tie a to b to c
16:07
to z
16:08
on the fly. It was something prebuilt out
16:11
of the box, health plan in a box,
16:13
if you will. And, that was the number
16:16
one thing that we we created. And
16:19
the reason we created it was not because
16:22
that helped us implement, because we wanted to
16:24
be a systems integrator to implement it. What
16:27
we realized that the true
16:29
need of a health plan wasn't new technology.
16:33
It was the guarantee
16:35
of what the actual operating cost of their
16:38
back office would be. Mhmm. Meaning, at a
16:42
literally a per member per month cost
16:45
that the health plan could take to the
16:47
bank and know that in order to be
16:50
at 10,000 members, a 100,000 members, or a
16:53
million members,
16:55
my price point for operating
16:57
process health insurance, claims, enrollment, provider, all that
17:00
kind of stuff,
17:01
was going to be at a very fixed
17:03
price guaranteed
17:05
from the day you sign on the dotted
17:07
line. So that when you think about that,
17:09
that is an actual business process
17:13
outsourcing,
17:14
not BPO. It's the actual process that you're
17:17
outsourcing.
17:18
And therefore, we said that that was a
17:20
business process as a service as opposed to
17:23
BPO.
17:24
Mhmm. And so when when a customer comes
17:26
to us and says, I need my operating
17:28
cost to be $10 PMPM.
17:31
10 all in whatever all in is, and
17:33
that can vary from customer to customer. But
17:36
and they say all in. That means I'm
17:38
going to give you responsibility
17:40
for
17:41
all of the technology
17:42
to execute that business process
17:44
and all of the resources needed to triage
17:47
any
17:48
fallout in that business process. And at the
17:51
end of the day, that's gonna cost me
17:53
$10 per member per member.
17:55
And that is a true outcome of pricing
17:59
that the health plan can then budget around
18:02
for a end period of time. And our
18:04
contracts are between five and ten years. So
18:07
some some customers know for the next five
18:09
years what their costs are gonna be. And,
18:11
obviously, there's tearing. If they grow from from
18:13
a 100,000 to 200,000,
18:15
the the volume
18:17
will lower the overall PMPM. But the the
18:20
key part of it is that the CFOs
18:22
can can understand
18:24
exactly where,
18:26
their costs are gonna be, and the COOs
18:29
have an SLA, a long list of SLAs
18:33
that define the performance and the quality of
18:35
what they're going to receive
18:37
for that business process.
18:39
So
18:40
it just to kind of sum it all
18:41
up, you know, BPaaS
18:43
is a combination
18:45
of the software and services needed
18:48
to run the actual business process
18:51
at a defined outcome price that is guaranteed
18:54
through the contract of of the deal.
18:57
I'm not an architect, but building the foundation
19:00
is arguably the most important thing for building
19:03
a house.
19:04
I would argue so much that I know.
19:06
Oh, Kevin, I I wanna close this out,
19:08
with a little bit of a future outlook,
19:09
and and you've touched on AI earlier in
19:11
our conversation too, and it's certainly something that
19:13
we're seeing in the space as well. And
19:15
we're now sort of in this process of
19:16
transitioning. Right? We have
19:18
generative AI gen AI, but now we're switching
19:20
to this the the agentic AI piece that's
19:23
really evolving, has evolved in 2025, and will
19:25
certainly evolve in 2026.
19:28
I'm wondering how you're seeing this affect
19:31
core administrative functions in health care and also
19:34
what it can mean for the future of
19:36
payer operations.
19:38
Yeah.
19:39
So I I will be completely honest with
19:40
you is that for the last
19:43
10 now let's prior to
19:46
I would say between 2016
19:48
and twenty twenty two ish,
19:51
I was an AI.
19:53
I did not believe in it. Yeah. I
19:55
didn't believe in it because there was no
19:57
proof points.
19:59
And I think the whole,
20:02
OpenAI
20:03
and and generative
20:05
activities
20:06
have opened up a whole new realm of
20:09
capability.
20:10
I will still say, as of today,
20:14
2025,
20:16
there is still limited
20:18
value points that have been realized. I I
20:21
can go into where we have seen the
20:23
value in health plans. It's not so much
20:25
agentic as it is ML,
20:27
Yeah. But, there's definitely areas of of focus
20:31
and and opportunity.
20:33
But I what I will say is that
20:35
the normal approach
20:37
to agentic AI that I see most
20:40
either service providers,
20:42
meaning SIs or or or people who wanna
20:46
implement it,
20:47
are approaching it as is essentially a glorified
20:50
version of RPA,
20:51
robotic process automation.
20:54
And you're using
20:55
AI to help you with the gathering of
20:58
data and interpretation of that data so that
21:00
you can execute a defined process to move
21:03
forward with.
21:04
I I find that to be
21:07
analogous to mess for less. You're you're you're
21:09
you still have a mess underlying you, and
21:11
what you're trying to do is put a
21:13
fix on top. You know, from my perspective
21:16
is that if you to put RPA on
21:18
top of a mess, you're actually managing another
21:21
layer of technology on the complicated
21:24
technology that you have.
21:26
You know, honestly,
21:28
prior to the combination of of HealthEdge and
21:31
and Healthproof,
21:32
our only alternative really would have been to
21:35
take that approach because there we didn't own
21:38
the underlying software.
21:40
As the owners now of the underlying
21:43
software that runs the back office of health
21:45
plans with the combination of health edge and
21:47
health proof,
21:49
you you have the ability to build a
21:51
Gentec within the core software itself
21:55
and to be able to,
21:57
understand
21:58
that,
21:59
as transactions are coming through the system,
22:02
our people remember because we are administering the
22:05
the health plans also.
22:07
Our people are looking at that data and
22:09
are able to interpret where
22:12
there are inefficiencies, meaning I do the same
22:14
thing every single day. I wake up in
22:15
the morning, I hit next next, next, next.
22:18
There's no real value in the human executing
22:21
that function. It just is, you know, the
22:23
human can correlate a to b to c
22:26
and be able to hit to say whether
22:27
it's right or wrong. Yeah. That knowledge now
22:30
can be conveyed to our product organization who
22:31
can then use other agentic
22:33
product organization
22:34
who can then use other agent capabilities
22:37
to build it directly into the software
22:40
such that the the next step in the
22:43
evolution will be
22:44
the answer is there, and the the agent
22:48
is validating that answer
22:50
for some period of time, let's say, a
22:51
month or two months. And then at the
22:53
end of that two months, it becomes an
22:55
automated activity within the software itself because
22:59
the the the enough proof points have been
23:01
done to train the engine to understand that
23:03
I no longer need to even present this
23:05
as a as an option.
23:07
And so,
23:08
you know, to me,
23:09
the best people to leverage
23:12
in the the notion that we're gonna move
23:14
towards is human less claims processing.
23:18
Right? The idea that it will be the
23:20
exception to the rule for there to be
23:23
a human in the loop of claims processing,
23:26
the only way we'll get there is through
23:28
leveraging adjunctive capabilities to automate all the activities
23:32
in health insurance.
23:34
Health insurance
23:35
problems
23:37
inherently come from poor data.
23:40
And,
23:41
you know, health insurance is an incredibly dirty
23:44
transaction.
23:45
Dirty in that there's lots of errors, lots
23:48
of nuances,
23:49
and lots of alternatives
23:50
to to how how data come, like, the
23:53
an enrollment transaction is called an eight thirty
23:55
four, and you'll always hear every health plan
23:58
say,
23:58
if you've seen 1834,
24:00
you've seen one eight thirty four. Right? They're
24:02
always going to be different and and done
24:04
differently. And so you're gonna use AI over
24:07
time to be able to normalize
24:11
the transactions
24:12
into an understandable
24:14
transaction that can then be automated
24:17
through through this solution.
24:19
Yeah. Again,
24:21
you come back to understanding people need to
24:23
understand what it actually does. Is it actual
24:26
AI? Is it automation? What does it do?
24:28
Again, coming back to the foundational piece that
24:30
we've touched on earlier. Earlier. Kevin, what a
24:32
wonderful conversation. Thanks so much for being here
24:34
and taking some time for us. This was
24:35
great.
24:36
It was my pleasure. And, Lucas, I hope
24:39
to to talk again. I I have,
24:41
as I said, twenty seven years of experience,
24:44
and and, and and I'm still passionate about
24:47
this business, and I really look forward to
24:49
helping make America healthier again. So Yeah. It's
24:52
so much to talk about. I I agree.
24:54
So many different topics. We certainly want to
24:56
thank our podcast sponsor Health Edge for this
24:59
conversation. It was fantastic.
25:00
And you can tune in to more podcasts
25:02
from Becker's Healthcare visiting our podcast page at
25:04
beckershospitalreview.com.