PegaWorld 2025: Vision for Healthcare in an Agentic World

PegaWorld 2025: Vision for Healthcare in an Agentic World

So. Good afternoon. My name is Robert Connely. I'm the global market leader of Healthcare for Pegasystems. Often a person they would put after lunch to do a presentation. You'll find out why in a few minutes. Um, try to keep everybody awake and up going. Uh, but anyway, today's conversation we're about to have is about Agentic AI as it is at PegaWorld this year, of course, but also its impact on health care. This is an industry I've been in for, well, too many decades, I suppose.

And it's I think one of the bigger industries that are going to have the greatest impact. Um, as far as what Agentic AI is able to do for that particular industry. And that's what I want to spend most of the time talking about today, not necessarily deep into the technology of AI, although the if the conversations take us there, then so be it. That would be great.

But I think one of the interesting parts is focusing on the business of healthcare and what's happening there that makes it so ripe for a genetic eye, and why I think that the platform that we've created at Pega may have one of the best opportunities to take not advantage, but improving some of the problems that we have in that particular area. So let's start out with the business of healthcare. I mean, what is healthcare? Well, in the US specifically right now, let's just talk about that.

We're pretty much in an unsustainable situation, and we have been for decades. I mean, the reality is, is that we spend more on healthcare and we get less. I'm not trying to be, you know, really weird or critical about it, but let's face it, we have an aging population. We have unsustainable chronic illness everywhere. We have inequity. We have all sorts of these factors that get in there.

Um, and this is what's happening is that these social factors, basically aging population, lack of caregivers, disease hospitalizations are starting to overwhelm. The American health care system has been doing it for years. It's now increasing it even more. And what this is doing is basically driving these unsustainable costs on the payers, on the government, on every one of us as individual citizens. So this has been the pattern now for a number of decades.

Ever since I was a child, this has been something we've been looking at as a problem coming downhill. But here's the issue. In healthcare, we've almost always focused on this as a business problem, right? We've how do we make doctors smarter? How do we make payers more efficient? How do we make the system itself work better? Well, here's the basic problem. The problem is not a healthcare system problem. It's literally a human problem.

The fact is, is that most of our problems that show up in an emergency room, or show up in a hospital, or show up in some sort of physician's room somewhere, are due to other things, socioeconomic factors, the physical environment we live in, health behaviors. Healthcare itself only amounts to about 20% of that. So 80% happens somewhere else.

And this becomes the real focal point of healthcare going forward is this realization that we can have the perfect system, and perfect systems are great, but our problem is not with the system. So it would suggest that even if we fix that, we would still have these overarching problems. And so what this is doing is it's basically forcing the health care industry to adapt.

And we see this all the time, but we think that the biggest impact is going to be on the payers themselves, because they're the ones that basically started out life as being financing the health care movement in the United States. Then later on, we decided we wanted to make them gatekeepers through authorizations and whatnot. And so they've evolved in these back office processes. This is what they've been good at.

This is how they've made money by profiting on these administrative aspects of health care. But now we're starting to see this big change. We're starting to see them become more care partners, less about administrative details. But now it's almost shifting the business model. So instead of profiting on administrative efficiencies, they're now looking at profiting from improving health. That's really where the game now comes. Can we prevent utilization?

Can we improve care enough that we're not overwhelming our care providers. Um, this is where it changes that role as well to collaboration. Right now, it seems somewhat adversarial. If you talk to a provider about the healthcare payer relationship, I'm sure the payers feel the same way. But the reality is, is with this new pressure on the system, we're starting to see this opportunity for this to break down. And now providers and partners and payers working in more partnerships.

And what this will do, we believe, is drive these holistic member experiences that we've been talking since my childhood. So it's been around for a while. So but it's not just a matter of change. We can go in there and we can talk about this is the vision. This is where we've got to go. But every day this is what our clients work with. And I'm talking about the payer clients. Now. These are the areas that they spend all of their time focusing on.

So while we have these overarching macro pressures, we're also faced with these day to day problems that we're having to do. And now we throw into this mix of regulatory acceleration, you know, the new prior authorization rules, the, you know, no surprise act legislation, things like that. Quality performance with the guardrails changing in CMS, stars and others, were starting to see huge amounts of our clients starting to get beaten down by these things.

I mean, just the introduction of GLP one drugs has a huge impact on these organizations cost of care. These are one of the reasons that drive that up. So now, you know, basically chronic illnesses now account for 70% of our cost of care just because of chronic illness, because it's not managed well, it's not maintained well. And this is the effect of it. And then we have these other outlying factors. Member loyalty. It's never been easier for a member to change plans.

Um, providers stress you can't open up a paper in the US without reading about this. And obviously this is a big thing in the US and in the workflow workforce issues that we all have to deal with. These are the areas that I, especially Agentic AI is going to have an enormous impact on, simply because it's going to make it better. It's going to allow us to move the ball forward. So it's a time that I hear all the time about innovation. Um, it's a time that we can innovate.

It's a time we can look forward. And I guess I'm somewhat known as an innovator, having done it a bunch. And you realize that innovation is not a building a better mousetrap. It's not creating a better idea. It's really a fact of looking at a problem in a different way, you know, and then solving that problem as opposed to the one that you were tackling, sort of finding another path up the mountain instead of the one we chose.

And Healthcare, I think, really falls into this area because we look at healthcare as a clinical problem sometimes that doctors manage and control and patients are victims of that. We have our healthcare systems and whatnot, and the reality is it doesn't work that way. In fact, healthcare and this is a big challenge point to everybody since this is one of those kind of presentations, it's not really a system. It's everywhere. It's more of a fungus than it is a system. In fact, it's a community of providers and payers and services that only work together when something happens in somebody's life. It's along a continuum, a continuum. We don't follow a continuum. We know a little about. All we know is that everything that happens on that sine wave, going from left to right in a lifetime of a person is going to touch these people. It's going to drive their costs. It's going to determine what happened in the past. It's going to predict what happens in the future.

And yet we're blind to most of this stuff. So as we see that reality, this is how providers get involved. So if I'm going along my timeline and suddenly, you know, a lump appears on my throat, what am I going to do? Well, I'm going to go ahead and I'm going to engage a provider. At that point in time. They're going to diagnose, they're going to test, they're going to treat, they're going to do all those elements. And that creates a whole bunch of workflow. And then the payers get involved.

So this connection between all of these players in the game, Sterling create these massive workflows that are going across these things. So this is the reality. This is what a patient has to deal with every day when they're involved in a healthcare episode or a situation like cancer. And it's the way the systems work in Iraq. They're not planned, they're not orchestrated or organized the way they should be. They just they happen. So you show up in a doctor's office unseen.

Suddenly it kicks off a bunch of processes and it moves on. But here's the issue. We don't know how. It's not a mysterious problem at all. We know how to solve this. We've been solving this for decades. In fact, many of our clients here, providers and whatnot, have been doing this. What they do is they use care managers. They use care navigators, humans, much like our caregivers at home, working with these people and helping them avoid these problems.

You know, this idea that you get a paper cut, it's allowed to fester and you wind up in a hospital, and it costs the rest of us hundreds of thousands of dollars. But the problem with these systems is they just do not scale. And this is what we believe is driving this promise of agentic AI.

The fact that if we could interact with people at a different level, providing a lot more information and then acting on that in real time, not just when they show up, which is almost too late in a health care situation. And this is what we want to get to. So I'm not sure what I did there. Um, this is the promise of Agentic AI.

This idea that we could drop an agent, whether it be, I guess today a genetic I would be a, you know, a gen AI agent basically that listened and acted with systems around it is what it's supposed to do. This is what we believe it is. But the thing about Healthcare that's so critically important, and maybe a barrier to a lot of agentic AI stuff, is the fact that these processes do not occur in a vacuum.

In fact, they are highly complex situations which you guys deal with every day that we deal with every day. But most of the public is unaware of just how complex and difficult these workflows are. They don't just happen on their own either. They operate in silos, and this is one of the bigger problems that we see.

Um, I'm talking about the entire room right now, and I know most of you personally, so I know this is what you deal with on a daily basis is how do we make systems work that have to span across these silos, that have to work with people that we don't even communicate with regularly and create some sort of future? This is obviously a big challenge, it seems like, and it likely is. But we think this is where, you know, we're seeing this come together now.

So if we think about agentic AI in health care in this context, you know, today we think of a large language model based agent that would sitting out there working, running its models, trying to figure out what's happening. But it doesn't work well in these areas right here. They're blinded by the silos. They can't communicate across business units. They can't pull all this stuff together. So they're sort of at a and they're hindered by this. Quite frankly, they won't make it that well.

This is where I actually came to Pega on this one, because I was retired successfully and didn't think that I was going to come back into business. And I saw what Pega was doing, and it blew me away. Quite frankly, this notion of taking a case architecture and putting the data into that to use as the springboard was just amazing. That's where the difference is, is that we put this ability in the center of an ecosystem. And my ecosystem is not just the payers enterprise.

This is actually the ecosystem of healthcare, which is provider enterprises, payer enterprises, and the human enterprises as they manage their own daily lives. But this is the uniqueness about Pega. And so as we look across, um, this notion of taking this data and curating it into a case is just to me was just phenomenal. Because what it does is it allows what doctors can't do. They can't tell you what's good or bad, by the way.

They cannot, you know, you can't change a medical record once it's been recorded. You can't go back and update it and say, oh, that wasn't quite right. So it's going to be there forever. That future AI brain that's reading that stuff is not going to know that that wasn't right. That was just a human. It has to pick up on that. So this is the problem.

You've got to not only connect to these systems, but you've got to organize and prep this data, curate it so that it's literally complete, accurate, relevant, timely, those attributes of data that are absolutely required for AI to work without hallucinations and mass problems that it would have otherwise. This is the beauty of the underlying structure of Pega. Um, I don't know any other company that does this quite this way. But this was the magic to me.

The fact that you're you're preparing the data for downstage use. And if you ever hear Alan Trefler speak, he talks about this from his early days, that that's what he saw. Is that the way we were using information, the way we were using data, the way we were using computers was not planning for this future where computers will do the work and not humans. So this is one of the interesting aspects of of why Pega thinking here.

The second thing that allows agents to work in this environment so well is this workflow structure. The fact that once I have a case and I have a workflow, I can literally tell an agent, you know, what to do through its large language, English or whatever language that you want to be able to put there. Um, it's an amazing way to do it successfully.

So we when we look at it that way, we call it predictable, because what you're doing is you're narrowing down the context of that LLM to a very narrow slice of the universe, right? That part of that workflow that they've got to focus on. And what that does is it sets the context for the LLM so it doesn't go rushing off and creating all sorts of threads throughout the universe. that it might think that where this thing is going to wind up.

So this is the key to that predictability, that ability to look at that information within the context of a workflow with high quality data. Now we can make decisions that aren't quite like prompting ChatGPT for a question like Alan was doing this morning about, you know, a chess move. This is really about life and death in a lot of these cases in healthcare.

So we want to make very sure that the data is not only accurate, but it's also being used at the right place at the right time and the right moment to get the most value out of it. You'll hear a lot more repeating about value in healthcare being the use of the information in the data, not just the availability of it and the quality of it. So the other thing that's happening, and I know we have organizations here that are doing this quite well is in the case of agents and interoperability.

So in healthcare interoperability is a huge issue because we have electronic health records. We have clinical records and hospital systems. We have payer systems. we have an an amazing amount of systems that are distributed all over the place. And when everybody somebody makes that a reality, suddenly I'm having an illness or cancer. In my case, you're suddenly going across half a dozen of these systems.

You're trying to make a process across systems that don't talk to each other regularly, that don't speak the same language. Well, how do you do that? So we have these interoperability standards in health care that have evolved for a long time, but they haven't quite taken off yet. And I've always wondered why. And I think this is the key. The fact that humans have been using the interfaces up until now, except for simple results exchanges and simple mechanical exchanges.

The reality is, is that we've not been able to take advantage of them. It sounds simple, but it's quite true. The fact that we have fire exchanges, we don't use them because doctors would be required to tell their staff to go use those systems to pump information back and forth. Now, agents stepping in and changes the paradigm. Agents say I need to use these channels. So I'm going to use those effectively.

I'm going to continually put short back and forth and I'll show you how that might be done in a minute, but I think that's where the direction that this is going into. But you don't have to believe me, even though I think I'm honest. Um, Gartner really spends a lot of time on this. They call it their business orchestration and automating automation technology. And it's always good to look at the thought leaders and see, you know, are we on track? Are we following the same way?

And I think this is a case where it shows, to me at least, that with what we're doing at Pega. And if you look at the diagram that Gartner creates in the far left side, you realize it looks awfully a lot like our Center-out architecture.

In fact, it almost overlays it, because that is the notion to do this kind of work in an enterprise or in an enterprise like Healthcare, then you're going to have to have this in a place that it sees things happening, that it knows when to do, that it can connect to other things. This is where the interoperable interoperability piece works as well, is within this picture. So I like to use this. I don't think I'm stealing it. It's their stuff. I want to promote it because it's a great idea.

We just happen to be somewhat aligned with their thinking on that one. So now we're seeing this new world emerge where we're becoming care partners of the payers, the providers like that because they have their spot to play in the cycle. But outside of that, they don't really follow patients that well. Doctors are not necessarily responsible for you when you go out of the door of the doctor's office, unless you call back in and do something on your own, they're not going to follow you.

They don't have the resources for that. Payers, however, are starting to change the model. So what's happening in the payer world is this shift toward value based care. This notion that we should be paying payers to, you know, manage the money well but achieve better outcomes. Now, until now, the outcome has been that the service was rendered and I pay for it. It met the rules. In the future, it's going to be different.

It's going to be what are you doing to make sure that whatever you do actually affects that person's life in a positive way, not in a negative or irrelevant way. And so this is what we're seeing, is this notion of the payers starting to spend much more time in this continuity of care world, of how do we help patients as they shift across in their domains from disease to disease, place to place, things like that.

Can we interject ourselves in their lives so that we can identify a paper cut, and address it before it festers? This becomes the model of the future. So now let's imagine this future of agents. I'm going to walk you through a couple of scenarios here to show how it might work.

And to me, I look at it from the continuum because as I work with, you know, yourselves and other clients, it's this continual amount of workflows that we see happening all the time as it goes across, starting with the provider relationships. Well, if you're an insurer, you can't really have a business unless you have a provider network. That's sort of the basis of this. And so they have to go out and create one.

They they contract with them, they credential them, they get their networks built around these things. These are amazingly complex processes that take a lot of time and energy. But if we think about a genetic eye and where it's going to start inserting itself, what you're going to realize is that it's really automating is what it is. It's not so much a genetic now. It's just it's automation at another level, but it's the same thing.

So imagine, you know, in the future where an agent might orchestrate these contract workflows for a seller that's out there, or a broker that's trying to recruit providers. This would be a huge way to improve that and make sure that the providers are well above all these things. These insights that they might be given from the contract, there's a whole lot of abilities there, especially in areas like, amazingly enough, fraud, waste and abuse.

How we can identify providers early on that there may be a problem. This is actually happening in the tech world. We're getting systems out there that are doing that. Credentialing is another area. This is probably a huge opportunity in the payer space, simply because everybody credentials every provider in the same way over and over and over. It's the most replicated thing in the world.

Um, what if somebody could be smart and say, hey, I want to master that, and I want to help become a shared service? I think this is something of the future of healthcare and how genetic AI is going to change relationships. They're going to realize that everybody can't do everything themselves. I mean, most other industries have figured that out.

I think healthcare is going to get there very quickly, and it's going to look at how can we share things like credentialing and, you know, that sort of thing. Onboarding is something they'll still maintain, but those other two are really a factor in how they bring people on board. They would help themselves by just exiting out a lot of a lot of expense in their operation, just simply because of that. Then finally we get to the more advanced things like integration.

Again, this is going to be that cornerstone I talked about earlier once the integration goes into play. The idea that you could help a provider reduce their administrative overhead because of that, you know, like reducing the authorization turnaround time or reducing claim pins, things like that become incredibly important to these people. So now let's look at it from the other side. The same thing applies.

You know, the whole enrollment onboarding things, especially in American healthcare, things like Medicaid, you know, Medicaid, which has this, this period of enrollment, which is not an enrollment period. It's actually when your financial situations drop to a point, you become eligible. So basically what you do is you drop in and out of Medicaid continually.

For most people that are at the borderline, some people are always poor and they're always at the bottom, but a vast majority of them come in and out of the system. So if you're a payer supporting a medicaid population, you've got an immense expense. I'm just managing, you know, those particular aspects of, you know, getting people on board and, you know, bringing them onto the Medicaid environment for Medicare.

As you're probably well aware, those are annual enrollments, periods in which everybody could change if they wanted to. And so this is a big effort around the payers that, you know, service up the Medicare Advantage population around that. So just in those two areas alone, the impact of Agentic AI is going to be pretty significant. Onboarding the same process as we do that. And then we get to these interactive experiences, which I think are where Agentic really shines.

The first one's a simple one. Just if you're enrolling for the first time, you've been assessed and whatnot, you're going to get to a point where you want to find a doctor. In this case, it's a primary care doctor. That's going to be the first one, and you're going to have to use that network information that you picked up from that provider onboarding period and hope to goodness that it's maintained correctly.

And then that's going to get your insight as to how I find a doctor and then schedule an appointment and do those sorts of things in the meantime in the back end. All of these systems are put into place to make sure that that individual can find that doctor closest to their house and be able to set that whole thing up. So that simple process just triggers off so many underlying workflows just to make it happen correctly. It's an amazing, you know, sort of thing.

But then we get into the real heavy lifting stuff, the actual encounters themselves. So just the simple fact of going and making an appointment with a doctor, scheduling the doctor and going, having that visit and let's say, God forbid, something he discovers or she discovers in that moment, you know, leads to maybe a prescription of something. Maybe you have pre-diabetes. Let's not create that. But that could be the case.

In which case what you see is an enormous amount of interactions that take place from planning the visit, scheduling the visit, doing eligibility checks, doing all those things, and then you get to the doctor, then you have to go through the testing, the examination, the diagnosis. You go to the planning phase. All of these things just create these massive workflows that trigger off of that one little incident that happens.

And this is where a lot of the pain comes from is all these little underlying components, people shifting around work, trying to find out what data is in the back end system. Are they eligible? Does it meet their plan requirements? Primary care is not as difficult as specialty care. Specialty care. It goes into a whole nother level of criteria.

Um, and the reason this little phase is so important is not just because it's there, but because if you're a provider, you're in the business of having people come visit you so that you can make money, you know? And I was president for dental company for a number of years, and it's button seats is what we called it. That was the term. How do you keep people coming in that you can actually create a business out of what we believe in is in the near future.

We're starting to see some of this now is that when somebody cancels something, a genetic eye in the background will start searching basically frequent flier list to start finding people and start marketing to them to see if they can fill that gap in that physician's day. That's a huge benefit to a provider when you can start managing the flow of patients, because you can see a bigger picture than you would have otherwise.

So this is one of those areas that we see Agentic AI really impacting health care. And then um, one we'll talk about quite often around here is one of the big ones, which is prior authorizations right now. So if we're familiar with the new P prior authorization rule 057, it sounds like a James Bond movie, but it's there and it's the one that basically requires individuals.

Um, well, not individuals, but payers to turn around prior authorizations and in some cases 72 hours changing the complete rules about what is required to be done, what's required not to be done, especially with denial notifications and things like that. And it's an incredibly complex process. Um, if you think about it from the provider's side, providers spend about 30% of their time in an American practice doing prior authorizations and battling appeals. That's what they spend.

They're not practicing medicine. They're not working on patients. They're actually doing this administrative criteria. In the meantime, on the back end of the payers, it's just as bad. I mean, you know, many of our large payers spend hundreds of millions of dollars a year in this particular area of I get a request for an authorization for a treatment.

I've got to see if that meets the plan, makes medical necessity criteria, all of these things that come into play to see if we should do that or not. While the provider waits, while the patient waits, while anger starts building up. So it's one of these complex processes that really are one of the things that I think New York Times, Wall Street Journal, I think there's a day goes by they don't have an article about this particular problem. So it's that kind of thing that people talk about.

Then we move into the next area. So now I've seen you, I've created these things. I now want you to go get medication. I need to get my exam paid for. Now the claim comes in. This is one of those other interesting problems, that the silo effect becomes so problematic because if I did a prior authorization for somebody, I have a lot of data that I use that prior authorization from.

Chances are, in most other organizations, when that claim comes in a few weeks later, they're not going to make that match. They're not going to have all that data. They're not going to be able to pull that together. So it's going to get what we call pended. I'm going to give somebody else to go chase this information down, see if they can find it. If they can't find it within a period of time, then they're going to basically deny it. That's going to turn into appeal.

The appeal process is going to go forward, and they're going to lose the appeal as far as the payer, because it was just a matter of lacking data. So these are the areas that we see agentic AI because it sees that continuum, you know, from the prior authorization from the provider, onboarding all the way through that claims moment. Suddenly you see a different picture. Now you're starting to do things like pre adjudicating a claim that's technically what's happening here.

And from that pre adjudication point we're able to reduce denials which is the provider's number one issue with the payer by the way. They seem to go in two directions right? Claim denials and then prior authorization denials. Those are the two that spend the most time in angst. And these are the two areas that I has a really interesting chance of doing this.

In fact, we're doing this now in a number of our clients that are using AI to look at a broader context as information flows into a claim area, for example, to make sure that we're touching all these points and that's working correctly. Um, I mentioned appeals. The appeal process goes on. The interesting thing about these appeals and all these other things is they go through there is to the patient, to that member, this is their life. This is not just a delay. This is not an administrative delay.

It's your life. And like I said, I went through that myself personally. And it's not good on you and your family when you don't know certain things. That unknown quantity of life is pretty problematic, and a lot of it occurs because of these delays that are happening in these very complex systems. We're not just not turning around, we're not letting anybody know what's going on, which is even bigger.

So if you look at some of the pushback that happens, you know, in the press right now, a lot of it's going toward that particular angle is we don't know what's going on. They're not letting us know. And all of a sudden we get a denial. Months later, we got to start over. This is a problem for the industry, but it's a problem that we believe that I can address because this is really just a lack of awareness, a lack of continuity in the data streams as it goes across these various business cycles.

Moving on down to the more complicated things like clinical areas like hospital discharge, these are incredibly complex areas that often involve multiple providers, multiple hospitals, multiple payers in some cases. This is where it gets really complex. But again, because of all that approach that we were doing upstream, it makes these even much more able to be done effectively. In the case of a discharge, you don't want them to be readmitted. That's something that's a huge problem for everybody.

So by using this agentic AI to interact with people in a different way, bringing some of these marketing technologies like CDH to play, which is an AI component, it becomes very, very capable of reducing these cost of care that I talked about in the very earliest segment. And then finally it moves on to the ongoing life of people.

How do we use Agentic AI in these high touch interactions that we can have with them to learn more about them, to understand what they're doing, to identify those paper cuts earlier? So this is the magic of Agentic AI in Pega in my mind, is all of these complex workflows can't be created by an LM brain. They really can't be done based on the previous architectures that we've had that basically aggregate data and try to build things on top of low quality data. I'll just call it that.

Um, that's problematic. But we're seeing this world now becoming the way that Gartner recommends, or at least leading toward this boat architecture. This is an example of it. Um, it just seems like the appropriate way to go. So you'll see more and more about this idea of using predictable agents, this idea of embedding them in a workflow and putting them together. There is just a phenomenal aspect of it, and we'll just go through these anyway. I know I'm about to run out of time.

I wanted to sort of give you a share of this and see if there were any questions. Um, comments. Criticisms are always liked, but, um, this is our world. This is the world that's changing very, very rapidly, um, even going into these other areas. So we Blueprint you see a lot of today, we've been doing some amazing work with Blueprint in the healthcare sector, especially when it comes to looking at broader context, because the workflows in healthcare are I think they're vastly more than most any other industry. You know, if we think about the number of sheer workflows, I mean, in most industries, there's thousands, tens of thousands and healthcare, there's tens of millions. So it's just an incredibly large problem to deal with.

But Blueprint is becoming an interesting thing to look at. These challenges that I just laid out and how could we plan for that? How can we think of that? How can we accelerate this change that we, most of us at least believe it's going to happen over the next ten years. And so this is the way we achieve scale to do that. These design agents are amazing.

I think I'm going to leave you with this one, because this is really where I think Agentic AI reaches us in something tangible that we can actually use and do. So not only can we build blueprints of functional workflows for agents, we can actually use them as discovery mechanisms and tools to look at a broad scope of work cases and use cases, and how we can apply them.

Now that we have CDH out there with the 1 to 1 blueprints adding to these journeys, it becomes even more powerful and capable of doing these sorts of things. Um, so that's my glimpse into the future. Um, a little bit. Are there any questions or things like that? I can't see? I've got klieg lights in my face, but it shines up my face, which always makes me nervous. So any questions comments that just stun you with us in this post-lunch sugar letdown? Could you step up to the microphone?

I've been asked to do that. Otherwise your voice is going to go silent and I don't want that to happen. That's right. Oh. Thank you. Well. So, uh, most of your use cases were around the administrative aspect of healthcare, and you didn't get to the care side of healthcare, right? In terms of AI informing the care plan, the diagnostics, the the execution, you know. So. Yep.

And the other so related to that, I would think an AI agent who mirrors a physician's assistant might be might be a potential use case. I think you're right. Absolutely. So I guess then the question becomes, well, who who's the master in the payer provider relationship? Right. Right now it's it's it's the payer. Well, I think that they probably think of themselves that way. Purse strings. Yeah, that's that's exactly right. Provider has to convince the payer to pay them. Yeah.

That's that's exactly right. That's the way it's been to me. That's on its head. The I would agree. The the provider should be the one who's the master of that relationship and dictating the process by which you get care. So anyway, that's just. No, I would agree and I think. Are there use cases that Pega. Yes. On the care side. Yeah. And so I wish I'd had more time. I've added that slide into it.

So what it is, it's basically looking at the member experience, not so much what the provider does but the member. So think of things like patient activation, medication adherence. Those types of things are where we see the workflows of Agentic I. But it'll actually be likely the payer that will have that technology that they'll provide. The providers will still do their stuff with their EHR. So you mentioned ambient AI as an example.

I believe, when you were talking about the provider and the, you know, the staff and how they're analyzing that sort of stuff, that's a perfect tool for that in the EHR. We've tended to avoid the clinical decision making because you're right, that's a provider. I mean, that's the provider world. The problem breaks down, though, when you're dealing with multiple providers. Then there is nobody in charge. You can say the primary care is involved and they're the one in charge until you get cancer care, and then they're no longer the primary care or you get heart disease. Now it's the cardiologists. And the oncologists are now leading that with the primary care involved. This is where it gets a little sticky. So I completely agree that the trust is with the provider. I don't think anybody's trying to change that. I think in many ways the payer will become more of a silent partner.

You know, a technology partner that the provider still has the front end line. But when it comes to continuity of care, that starts breaking down a bit because there is no provider that's responsible for that one. I know because I just went through this myself, and you're kind of on your own when you're between multiple providers because they don't know exactly what the others were doing. They don't know. I mean, we share continuity care plans.

We do all sorts of technical things, but for the most part that's difficult. So completely agree with you. I'm not trying to say that we aren't focused more on the low hanging fruit of administrative things at this point in time, and you're absolutely right. It's headed that way. I don't know that we'll be leading that one so much. I think the EHR vendors are really taking a big forward leaning motion when it comes to things like ambient.

I just what I think of when I think of that provider use of it, we're not going to get into clinical decision making. We're not going to get into that sort of thing. That is the providers world. But I think what the payers are going to do is they're going to become more of a trusted partner. If they can pull that off and be that for those providers, I think we're seeing it in the provider world. They become much more aligned that way. Does that help? Yeah, that's very helpful. Thank you.

Any other questions? And we'll be around all week. Stop by. Um. Is this a countdown? It is a countdown clock. So clock? So I got ten more minutes. I should have just had more. Yeah. Oh. I'm sorry. Please. In the use of agent and agents and, you know, as part of a process or workflow that has to be accomplished, how do we deal with the compliance part of like Healthcare when we have semi-autonomous or autonomous agent workflow in place, how do we look at the compliance aspects?

Yeah, I think that's a perfect question about compliance when it comes to how to use agents. I think in many ways, Alan addressed it this morning. You can't leave compliance up to agents. You know, that has to be sort of built. They have to be built in a protected room. So when they make their decisions that you're not violating a compliance issue, um, especially making decisions around something that that may violate regulatory sort of issue that they might have.

We see these sort of things a bit. So what we're trying to do is say compliance. You have to manage. It's part of your criteria when you make decisions, the problems with a large language model that you don't know all the sources that it's drawing from to create its patterns and its context. This is why we believe more in that when it gets to data, it's almost this retrieval, augmented generation, or Rag models that people are tending to do.

So we will limit the amount of data that the LLM can actually use in its decisioning. That helps with compliance especially. I know I've got my partner here sitting in the front. We spend a lot of time with this. How do we teach it? The rules of the road? Or at least understand those rules so that it's baked into the decisions that it makes, and not just hope that it assumes that, you know, in its in its response when it comes back from your prompt. So you'll find this to be prompt.

Engineering will be the future of this one. This is what we're having to do is say you don't know the rules. So we're going to have to teach you the rules, you know, as part of the prompts initially, and then maybe it will become more of an object there. So that's what's happening.

We're also seeing a move back toward the smaller LMS that people are doing internally themselves to lower that risk so they can be more in control of what compliance rules and other things that they're making sure are being followed. That's the trick. This is why we talk about Pega being so good for the agentic spaces, because these are the things we start out thinking about. And then you can inject AI in that in some limited fashion at first, and you can expand it later.

Um, hopefully long before we get to the, you know, the world of AGI and the singularity we were talking about earlier. Um, I don't think that's part of our conversation. If that happens, we won't know it happened probably as the answer to that question. So any other questions? Yes, please. So, um, this whole conversation around AI, in terms of the, for instance, the prior auto education process and stuff like that.

I think some of the questions have been about kind of the path to get there, right. And as you know as well as I do, that there's a lot of regulatory pressure right now and actually legal pressure around when we can use AI in the prior auth process, even in very large health plan has been accused of dialing it up or down for profit margin.

So I think that when we think about the vision of being able to get to some of these different places, even in the administrative piece of really getting to its full functionality, we're going to have to just be patient as it works through the regulatory oversight and the directing of the CMS and the other bodies. Yeah, because there's this propensity by the plan, by the payers to try to get profitability. Are you speaking to all of us or just because we can't hear you? Oh, you can't hear me?

I'm sorry. I don't mean to. Yeah. You got to talk to the microphone. I'm usually pretty loud. I'm sorry. I'll just recap. What I was saying was, is the vision of being able to use a lot of this AI technology to influence the automation and the decision making around prior auth. And some of the other visions that is being presented is great. The problem is, is you have CMS oversight and regulatory pressure right now pushing back as to what decisions can be automated.

So to your point about prompting the right decision maker is very valuable, right? So you have to have that oversight. But there is this undercurrent of we can't move too fast as an industry because we we end up getting into stubbing our toes and having a higher, higher focus on profitability as a payer. I like rubber. I work with payers all day long, so we have this propensity as payers to try to make sure we're managing our profit margins, especially with Medicare right now being so tight.

So there is a lot of regulatory pushback and legal pushback about how fast we can achieve the vision, just to give context. That was my only point. No, absolutely. And you're right that a lot of the pushback is happening. They're using, um, I don't know if everybody's familiar with these problems that are happening right now, but they usually make them to make approval and denial decisions.

Um, and that's a huge problem that people are seeing that, you know, you can't just give it up to an agent to make a decision that a doctor would do, even though it may comply with all of the medical necessity and all the rules around that. We don't even suggest that at Pega we're like, okay, the decision is either made by a statistical model. So it's not a genetic in the sense that it's not, you know, a symbolic large language decision. Um, and then it will be fenced in by all those other aspects of it. So it's not like we're ever going to issue a denial for a claim, which is, I think the number one thing that happens because it didn't fit a threshold, um, they will use AI to get up to that point where a human makes that decision. Don't get me wrong, but it's still a human in the middle. Um, we see in these things. But a great point. I don't know if that was valuable or not. It was a great, great, great all these conversations.

And we've got to be talking about this stuff for a while. Guys, this is not something that's going to go overnight, you know, to the next phase. I think half of you in here are in the beginning of the leading edge of this wave. So I feel like I'm talking to people that should be educating me. But, you know, I think this is a point that we all have to take out to the world, though. This is stuff that, you know, it's our responsibility to do to explain this to them, you know? Yeah.

You're afraid about AI. Nobody understands it. Well, well, I think we're overblowing a lot of what we don't understand. We're imagining things, but we always do. I mean, God knows computing started this way. I can recall that, you know, it was going to take over the world and was going to kill us all. Um, and I think AI is probably in that same boat, but all of it maybe has some truth. I don't think so yet. But the fact is, it's changing our lives. It's changing it now.

I mean, I don't imagine anybody writes a document anymore to start with. They just. Well, I'm going to write a prompt and see what that document looks like. Um, so it's already affected our lives. Now we've got to turn it into a force for good. And I think healthcare is that greatest opportunity to do that, you know, to see where we can apply this in such a way that we actually get benefit from it, you know, without just the entertainment value, which is great. Don't get me wrong, I love it.

But that's not going to solve our problem. It's not going to fix cancer. Um, come up with some great ideas to go about it, but you know, it's not going to do it. I thank you, everybody for listening to me. Um, have a great one. Enjoy. PegaWorld.