Historically, Healthcare Innovation has focused on health system tech deployments with less focus on what is happening on the payer side, although over time the rise of “payviders” and integrated health systems has shifted that focus somewhat. But last week I had a fascinating conversation with Kevin Adams, CEO of HealthEdge, one of the largest providers of software and services to the health plan market. One of the things he said that surprised me is that his company usually does not engage with health plan CIOs, but works directly with CFOs and chief operating officers on business process transformation. Read on for a lightly edited version of our conversation. 

First, some background about Adams: Among other positions, he has served as CIO at Cigna and CTO at Edifex. In 2015, he founded a company called UST HealthProof. Here is his description of that launch: “I wanted to start a business that was focused on community and regional health plans to help them become more efficient and effective, so that they could better compete with the nationals,” Adams said. “Most of our customers were provider-led health plans that needed a significant amount of operational efficiency, so we built an entire model that combined technology and the services to run the business on top of the technology and provided a guaranteed outcome, meaning claims were processed, and it was a guaranteed price, and that price typically was 20% to 40% cheaper than what anybody else could do. We took it from zero to $520 million in revenue and 5,000 employees over the course of 10 years. Then Bain came and bought us and said you're going to go help everybody, not just the small guys.”

Indeed, in 2025 Bain Capital acquired UST HealthProof for $1.3 billion and merged it with portfolio company HealthEdge. 

Healthcare Innovation: Could you start by talking about that merger of UST HealthProof and HealthEdge?

Adams: UST HealthProof was HealthEdge's largest customer. We were an important part of the HealthEdge business. Most integrators say we'll integrate anybody's stuff. But because our drive was to simplify and reduce cost, HealthEdge was the backbone of our approach, and there are lots of benefits to that in terms of focus. 

We became a very large customer and could drive a lot of cost reduction in the software pricing, but more importantly for us, all of our processing, all of our implementation resources, everybody only had to know one piece of software. We didn't have to know the myriad of things that most integrators like an Accenture would need, so we could be much more efficient, much more effective, because we were so focused on what we did. 

We call our business model “business process as a service.” We went to health plans and said we will process your claims with accuracy and compliance and all those basics, but we're going to do it 30% or 40% cheaper than you can do it. The only thing health plans today can do to be more efficient is to work on their operating costs. The things around medical loss ratio [MLR]  or the medical side of their cost equation are going to take years to get fixed through either renegotiations or premium increases.

HCI: Do a lot of health plans have internal systems that they try to develop and maintain themselves or do most of them work with a main vendor partner, much like on the provider side there's a big EHR vendor like Epic or Oracle?  

Adams: Today, if you break the market up, there are two large suppliers of what I will call modern technology, and one of them is us and the other is the Trizetto arm of Cognizant. Cognizant has maybe an 8% larger market share, but we're both in that roughly 20% range of market, so occupying about 40% of the market. The other 60% of the market is made up of either legacy mainframe solutions that were typically homegrown or legacy small software providers that are not improving their software anymore, but it's too difficult to move off, so people are still on it. There are a number of small guys out there, but the largest number is on mainframes still.

HCI: And is that problematic? With the way that technology is changing and AI is being brought into things, are the plans that are still on those legacy mainframe systems struggling?

Adams: Yes, so this gets back to my comment around how Bain is going to make us service everybody. It is actually both really good for our business and equally challenging, because the mainframes have historically been very large. If you look at Anthem's mainframe, it's got 35 million members on it; Cigna’s mainframe has 20 million; Aetna’s has 26 million. Historically there hasn't been modern technology to support that degree of size. But we just finished a validation with AWS to scale our product to 35 million members, so we now know that we can support that level of membership, and therefore health plans are coming to us in droves. The ability to support mainframes is getting very difficult because of aging out, and nobody can do COBOL anymore, not to mention the IBM stuff is very, very expensive on the mainframe side to support.  But most importantly, the lack of flexibility in that software — I mean, these are 1970s-era, so the amount of complexity and the lack of flexibility is forcing these health plans to move. I can't tell you the name, but we are in the beginning of a very, very large mainframe population move, and we're super-excited about what that means.

Also, CMS has initiated an RFP to move all of fee-for-service Medicare, which is 34 million Americans. They are going to do a bake-off over the next six months to remove or replace the mainframes that run fee for service.

HCI: You held the CIO role at Cigna. What issues are the health plan CIOs grappling with today? What keeps them up at night?

Adams: Honestly, our business model doesn't really talk to the CIO all that much. Typically when you look at a health plan, the leadership focuses on product innovation — unique reimbursements, unique networks, unique benefits, unique pairings of a variety of different capabilities. The second thing that they focus on is cost, whether it's driving down medical loss ratio, which is 85% of every dollar, or the administrative loss ratio, which is 15% of the total dollar spend. 

HCI: OK, let’s shift a little bit and talk about AI. Have the health plans been fairly early adopters of AI or have they been laggards? Or somewhere in the middle?

Adams: I think everybody has tried to do something. The pixie dust is a very hard thing to walk away from. I was recently talking to one CEO who has a board directive to have an amount of savings driven by AI, because the board believes that it could happen.

You have two forms of AI, and today 95% of what's happening is in the first form, and that first form essentially ends up being a glorified version of RPA, or robotic process automation, where you don't change the underlying problem but you put a wrapper layer around it to address the problem. What I'm saying to most anybody who will listen is that you now have to maintain two complex systems. You’ve got a whole group of people who understand how to maintain this wrapper layer, but then you still have this underlying legacy technology footprint. So as long as you're having to maintain both of those, you're never going to get efficiency out of it.

One of the reasons why I was so eager about the combination of HealthEdge and HealthProof is that the HealthProof side of the equation are business people who understand how to make the claims better, faster, cheaper, while the HealthEdge side have the tech savvy. Without those two things together and fully aligned, it was very difficult to get the solution to where it needed to be.

Our goal as HealthEdge is to embed AI directly into the core processing engine to eliminate the problems from ever happening in the first place. 

HealthProof’s drive was to get costs down as low as possible, and we're probably one-fifth of what it was when we started the business 10 years ago. I want to get that cost as close to nothing as we can get it using AI tooling, and we're already making great strides on that. Then I think over the next two to three years we'll probably eliminate, 50% of the people needed to process claims by using AI embedded directly into our, products.

I believe 95% of AI today is not driving measurable outcomes, but the 5% is driving huge outcomes. When you’re actually using AI at the core, then you're able to achieve much better outcomes. The problem that I have with AI right now is the the token cost is so high that when you start to embed this stuff into claims processing, it becomes an uncontrollable token cost and the way the systems are built today is that they're they're still fairly monolithic, and they're not broken down enough to be able to isolate your token usage to a particular piece of the business. We have some work to do on the software itself to enable it to be leveraged, and then the second thing is until token cost can get down from 40 cents to 5 cents, it’s going to be very difficult to really make it ubiquitous.

HCI: When we interview people on the provider side, we hear a lot about  a competition between providers and payers using AI around prior authorization and claim denials. Do you see that? 

Adams: One of the benefits of Bain Capital buying us is they own athenahealth, so I specifically requested a connectivity between us. They went so far as to put their chief product and officer, Paul Brient, on our board. He and I are actively discussing ways to bring these two capabilities together. His technology creates a claim, my technology processes the claim, so why does that have to be so far apart? In the first conversation I had with Paul, I said to him we are starting a bot war in the UM [utilization management] field, and that will help nobody. Mostly through ambient AI, they determine that an authorization is needed. They automatically gather the information, create the authorization, and send it to the health plan. Most health plans today don't have AI responding. So they manually respond, and if it is denied, the rebuttal and appeal to that denial takes one second. So it's only a matter of time before the denial takes one second on the health plan side, and then the appeal will take one second, and then it turns into a grievance, and you're right back to where you were in the first place where you’ve got to have two people talking to figure it all out.

I don't say this very often, but the government needs to establish some better rules for this at the Medicare level. That’s 70 million Americans — a little over a quarter of the total American population, so I think that's a great place to start. It’ll get a good proof point for how you can do it in the commercial markets, also. And that will be a dramatic cost improver, because it'll simplify the overall problem.