Building a Platform for the Agentic AI Era of Healthcare – Healthcare Innovation
Navigating the shift from volume to value in healthcare
The agent wars are here, says Penguin Ai CEO Fawad Butt. “It is not this futuristic thing that’s going to happen. It’s happening today.”
Penguin Ai, a startup that has built an AI-based platform designed to streamline administrative inefficiencies in healthcare, has pulled in $29.7 million in venture funding to pay for product development and to scale up deployments with payers and providers. Fawad Butt, founder and CEO of Penguin Ai and former chief data officer of UnitedHealthcare, Kaiser Permanente and Optum, recently spoke with Healthcare Innovation about the transition taking place to the new world of agentic AI.
The Palo Alto, Calif.-based company says its flagship platform combines task-specific small language models (SLMs), digital workers and agents, with a healthcare-specific AI platform to streamline processes such as prior authorizations, claims processing, medical records summarization, and appeals management.
Healthcare Innovation: Could you talk a little bit about your background as a chief data officer and how it helped you form the idea for this company?
Butt: Before starting Penguin, I spent a little bit of time in the VC world at Canvas Ventures, being an operating partner, and before that I was the chief data officer at Optum, responsible for all data and analytics capabilities across all three lines of business at Optum. Before that, I was on the insurance side of the house at United Healthcare as the chief data and analytics officer. Prior to that, I was in the same role at Kaiser Permanente. So when you have that unique vantage point, you see patterns. One of the patterns I saw was that the data is very disorganized. That’s why they created roles for for people like me.
The data is disorganized because over the course of multiple innovations, we bought best-of-breed solutions, and all of a sudden we’ve got a spaghetti mess of 10,000 systems trying to push data back and forth, so that was always a challenging thing.
Also, when I was in those positions, I could buy any platform in the industry, because we had the budgets to do it. I could go to AWS, Google Cloud or Azure, but the problem was that these are horizontal, generic platforms. They didn’t understand healthcare. If you bought that technology, you had to assemble a team and figure out how to use these generic tools and apply them toward healthcare problems. So to me, the thesis was always that healthcare deserves its own platforms, right? And why aren’t there more of those? And if I ever had the opportunity, I was going to build one.
HCI: Isn’t it also an issue that traditionally the payers had the claims data and the providers had the EHR data, and they didn’t really want to share?
Butt: The reality is that for our health system to operate, that data does get shared. A provider has to provide medical records if you’re submitting a claim or a prior authorization, so that data does get shared. The foundational argument that I’ve heard throughout my career is that the incentives are misaligned. I agree with that, but the reality is that the incentives are misaligned, and we share the data nevertheless.
The bigger challenge I see is that a lot of these companies have grown through acquisitions — United Healthcare, for example. Every time you purchase a plan, you get 13 core systems that come with it. If you do that 20 times, all of a sudden you have spaghetti systems doing too many things.
The reality is that there hasn’t been a technology evolution, until recently, that could take that spaghetti mess and in affordable way come up with a new architecture that could be deployed. But now the world is going agentic for the most part, which means the formats don’t matter anymore. Agents can pretty much understand and consume both structured and unstructured data and context around that.
So the next version of healthcare is agentic healthcare. If we believe that premise, and you’re a CIO or a CEO at a hospital system or a payer, you might say, I have to build 1,000 agents, right? You need a platform to be able to do that on. You could use the Azure platform or AWS and try to teach it healthcare. Or you can come to Penguin and talk to us, because we built the components that are required for an enterprise to adopt a platform. We have the connectors to your data, whether it’s in Epic or Cerner or athena or Oracle or Data Bricks or Snowflake — you name it. We have that built in. We have a privacy engine built in that de-identifies that information. We have a bias mitigation and scoring algorithm that checks to see if you’re introducing bias into your models. We built our own small language models for prior auth, risk adjustment, and claims adjudication, and then we give you our agents out of the box. That’s what a platform is supposed to do. It’s supposed to give you what you need so you can get to ROI in 90 to 120 days.
HCI: So are you envisioning both the health systems and payers as potential customers?
Butt: 100%. Because, to me, it’s a false dichotomy in many ways. Things start on the payer side and end on the provider side, or start on the provider side and end on the payer side. They just call them different things, right? One might be called claim scrubbing and one might be called claims adjudication, but it’s essentially a very similar processes. So why not build it once, and then let’s bring the data and the intelligence together in one place, so it’s not payer against provider, it’s payer and provider.
HCI: I was talking to a health system CIO who said they were starting to build agents in house, and the payers are building their agents. And it’s almost like it’s going to become this battle between the two to see whose are better. And she said that agent war is not where they want to be.
Butt: No, but that war has started. The agent wars are here, right? It is not this futuristic thing that’s going to happen. It’s happening today. I sat with the CEO of one of the largest regional health plans in the country. He said what they are seeing is that, in some ways, the providers have adopted agents a lot quicker than the payer side, because the payers’ processes are more complex. In one scenario, he said, a small network of providers that used to do 5% appeals on denials is now doing 100% appeals on every denial the health plan is sending them. He believes the provider group has an agent on their side, and the health plan has eight people on its side. So how are they going to win that?
HCI: Well, as part of your pitch to these execs, do they have to ditch a lot of this investment they have in all these expensive legacy tools that they’ve already made a bet on?
Butt: No, if you have a starting position that says, let’s call that sunk cost and start afresh, you haven’t been in healthcare long enough, right? That’s just not how it’s going to work. The investment in terms of technology might be legacy, but in terms of information, it is gold. All of the business’ processes are captured in data all across these enterprises. So if you are envisioning an agentic world, then you don’t look at it as old technology or new technology. You look at it as useful data or not useful data, right? And I can assure you that there is some very useful data in these legacy systems that is going to enrich the intelligence. If you’re building small language models or refining large language models, that information is going to be critical. We’ve been hearing that data is the new oil for a long time, but the honest truth is that now is the time where, if you can build the right refinery, you can turn that oil into fuel for your organization.
HCI: Does generative AI have potential to alleviate a lot of the problems that people describe about data standards or semantic interoperability?
Butt: I think it does. It’s already playing a major role. It used to be that the healthcare data world operated in structures or in data models. So I have a data model and you have a data model. If I need to consume your data or share my data with you, our data models have to be aligned by building a bridge in the middle which says your x means my y. So you spent a lot of time and energy building those bridges because that requires subject matter experts to weigh in, and those are hard to find and expensive to retain. But now because we’re not trying to normalize data before we can apply intelligence to it, we can consume that information in both formats and be able to analyze it in very similar ways and derive insights in very similar ways, so that step that used to take 18 to 24 months of just aligning the data…… I’m not going to say ETL [extract, transform, and load] is dead, but it is less relevant in the agentic world.
HCI: UPMC Enterprises is one of your investors. That seems like an organization that would be an obvious potential customer.
Butt: We hope so. We cannot comment on that. They’ve been a fantastic partner for us. They were one of the early ones to to appreciate what it is that we were doing, and they’ve been supportive consistently throughout the process. I would say in our earliest days, the institution that recognized what we were doing was UPMC.
HCI: Is part of what interested investors the team you’ve been able to build in terms of execs with a lot of experience in healthcare?
Butt: Yes, three of the technical leaders — head of AI, head of delivery, and head of engineering — are ex-Optum people. Many of the team that we have either worked at Kaiser, United, Optum, CVS and others on the technical side. Then we have our chief strategy officer, Mark Caron, who was the CIO at three different Blues plans and a CTO at Catholic Health Initiatives, as well as the CIO of the Collaborative Care Division at Optum.
And Missy Krasner, who’s been an advisor for us from the beginning. She’s been at Amazon, Google Health, and Redesign Health. I think that shows in the completeness of the vision and and it shows in the understanding of the nuance on governance and security and privacy that is particular to healthcare.
HCI: Are there some things on the policy and regulatory side, or involving marketplace pressures, that are giving you tailwinds to get more engagement?
Butt: I think there are two big tailwinds. One is that payers have been decimated over the course of the last 18 months. The Medicare Advantage business that used to be highly profitable has turned and their medical loss ratios have gone through the roof. So there’s massive pressure on them to reduce costs. And the cost that they can control is not the MLR, it’s the operating cost of running the business.
Then CMS has some regulations around interoperability and specifically around prior authorization that people are trying to recalibrate around. With an agentic world becoming reality, a lot of people are choosing agents to do that automation vs. another application that they used to do. So we are seeing a tremendous amount of demand and a number of engagements on that use case on the payer side. Risk adjustment is another use case on the payer side that is quite robust for us. Claims is developing as another area of of demand. On the provider side, I would say it’s around revenue cycle management — the claims scrubbing, the extraction of medical codes. Medical coding itself is a big use case for us. And then appeals is the other one. We’re going to be launching some additional capabilities that we’re piloting around patients and providers that will be coming down soon.
HCI: Do some agentic efficiency gains to health systems or payers come from reducing administrative employee head counts? Or are they seen as tools for those employees to use?
Butt: I don’t see any replacement of head count anytime soon, frankly, and it’s because the system is capacity-constrained right now. The nurses or doctors or admin staffers are being asked to do more and more. An example is prior authorizations. One of the big payers told us their pharmacists used to have to review 40 cases a day for prior authorization. Then that went up to 70 cases a day. Now they’re asking their pharmacists to do 100-plus case reviews a day. And guess what’s happening? They’re burning out. They’re making mistakes. They’re quitting and leaving the field. I have a personal story around this, because my wife used to be a prior authorization pharmacist, and I saw her go from working eight hours a day to 14 hours a day trying to make quota. So I think it’s a capacity issue. It’ll just relieve the people who have the responsibility to do this work to focus on the most important part of the work and work at the top of their license.
David Raths is a Contributing Senior Editor for Healthcare Innovation, focusing on clinical informatics, learning health systems and value-based care transformation. He has been interviewing health system CIOs and CMIOs since 2006.
Follow him on Twitter @DavidRaths
source
This article was autogenerated from a news feed from CDO TIMES selected high quality news and research sources. There was no editorial review conducted beyond that by CDO TIMES staff. Need help with any of the topics in our articles? Schedule your free CDO TIMES Tech Navigator call today to stay ahead of the curve and gain insider advantages to propel your business!
The agent wars are here, says Penguin Ai CEO Fawad Butt. “It is not this futuristic thing that’s going to happen. It’s happening today.”
Penguin Ai, a startup that has built an AI-based platform designed to streamline administrative inefficiencies in healthcare, has pulled in $29.7 million in venture funding to pay for product development and to scale up deployments with payers and providers. Fawad Butt, founder and CEO of Penguin Ai and former chief data officer of UnitedHealthcare, Kaiser Permanente and Optum, recently spoke with Healthcare Innovation about the transition taking place to the new world of agentic AI.
The Palo Alto, Calif.-based company says its flagship platform combines task-specific small language models (SLMs), digital workers and agents, with a healthcare-specific AI platform to streamline processes such as prior authorizations, claims processing, medical records summarization, and appeals management.
Healthcare Innovation: Could you talk a little bit about your background as a chief data officer and how it helped you form the idea for this company?
Butt: Before starting Penguin, I spent a little bit of time in the VC world at Canvas Ventures, being an operating partner, and before that I was the chief data officer at Optum, responsible for all data and analytics capabilities across all three lines of business at Optum. Before that, I was on the insurance side of the house at United Healthcare as the chief data and analytics officer. Prior to that, I was in the same role at Kaiser Permanente. So when you have that unique vantage point, you see patterns. One of the patterns I saw was that the data is very disorganized. That’s why they created roles for for people like me.
The data is disorganized because over the course of multiple innovations, we bought best-of-breed solutions, and all of a sudden we’ve got a spaghetti mess of 10,000 systems trying to push data back and forth, so that was always a challenging thing.
Also, when I was in those positions, I could buy any platform in the industry, because we had the budgets to do it. I could go to AWS, Google Cloud or Azure, but the problem was that these are horizontal, generic platforms. They didn’t understand healthcare. If you bought that technology, you had to assemble a team and figure out how to use these generic tools and apply them toward healthcare problems. So to me, the thesis was always that healthcare deserves its own platforms, right? And why aren’t there more of those? And if I ever had the opportunity, I was going to build one.
HCI: Isn’t it also an issue that traditionally the payers had the claims data and the providers had the EHR data, and they didn’t really want to share?
Butt: The reality is that for our health system to operate, that data does get shared. A provider has to provide medical records if you’re submitting a claim or a prior authorization, so that data does get shared. The foundational argument that I’ve heard throughout my career is that the incentives are misaligned. I agree with that, but the reality is that the incentives are misaligned, and we share the data nevertheless.
The bigger challenge I see is that a lot of these companies have grown through acquisitions — United Healthcare, for example. Every time you purchase a plan, you get 13 core systems that come with it. If you do that 20 times, all of a sudden you have spaghetti systems doing too many things.
The reality is that there hasn’t been a technology evolution, until recently, that could take that spaghetti mess and in affordable way come up with a new architecture that could be deployed. But now the world is going agentic for the most part, which means the formats don’t matter anymore. Agents can pretty much understand and consume both structured and unstructured data and context around that.
So the next version of healthcare is agentic healthcare. If we believe that premise, and you’re a CIO or a CEO at a hospital system or a payer, you might say, I have to build 1,000 agents, right? You need a platform to be able to do that on. You could use the Azure platform or AWS and try to teach it healthcare. Or you can come to Penguin and talk to us, because we built the components that are required for an enterprise to adopt a platform. We have the connectors to your data, whether it’s in Epic or Cerner or athena or Oracle or Data Bricks or Snowflake — you name it. We have that built in. We have a privacy engine built in that de-identifies that information. We have a bias mitigation and scoring algorithm that checks to see if you’re introducing bias into your models. We built our own small language models for prior auth, risk adjustment, and claims adjudication, and then we give you our agents out of the box. That’s what a platform is supposed to do. It’s supposed to give you what you need so you can get to ROI in 90 to 120 days.
HCI: So are you envisioning both the health systems and payers as potential customers?
Butt: 100%. Because, to me, it’s a false dichotomy in many ways. Things start on the payer side and end on the provider side, or start on the provider side and end on the payer side. They just call them different things, right? One might be called claim scrubbing and one might be called claims adjudication, but it’s essentially a very similar processes. So why not build it once, and then let’s bring the data and the intelligence together in one place, so it’s not payer against provider, it’s payer and provider.
HCI: I was talking to a health system CIO who said they were starting to build agents in house, and the payers are building their agents. And it’s almost like it’s going to become this battle between the two to see whose are better. And she said that agent war is not where they want to be.
Butt: No, but that war has started. The agent wars are here, right? It is not this futuristic thing that’s going to happen. It’s happening today. I sat with the CEO of one of the largest regional health plans in the country. He said what they are seeing is that, in some ways, the providers have adopted agents a lot quicker than the payer side, because the payers’ processes are more complex. In one scenario, he said, a small network of providers that used to do 5% appeals on denials is now doing 100% appeals on every denial the health plan is sending them. He believes the provider group has an agent on their side, and the health plan has eight people on its side. So how are they going to win that?
HCI: Well, as part of your pitch to these execs, do they have to ditch a lot of this investment they have in all these expensive legacy tools that they’ve already made a bet on?
Butt: No, if you have a starting position that says, let’s call that sunk cost and start afresh, you haven’t been in healthcare long enough, right? That’s just not how it’s going to work. The investment in terms of technology might be legacy, but in terms of information, it is gold. All of the business’ processes are captured in data all across these enterprises. So if you are envisioning an agentic world, then you don’t look at it as old technology or new technology. You look at it as useful data or not useful data, right? And I can assure you that there is some very useful data in these legacy systems that is going to enrich the intelligence. If you’re building small language models or refining large language models, that information is going to be critical. We’ve been hearing that data is the new oil for a long time, but the honest truth is that now is the time where, if you can build the right refinery, you can turn that oil into fuel for your organization.
HCI: Does generative AI have potential to alleviate a lot of the problems that people describe about data standards or semantic interoperability?
Butt: I think it does. It’s already playing a major role. It used to be that the healthcare data world operated in structures or in data models. So I have a data model and you have a data model. If I need to consume your data or share my data with you, our data models have to be aligned by building a bridge in the middle which says your x means my y. So you spent a lot of time and energy building those bridges because that requires subject matter experts to weigh in, and those are hard to find and expensive to retain. But now because we’re not trying to normalize data before we can apply intelligence to it, we can consume that information in both formats and be able to analyze it in very similar ways and derive insights in very similar ways, so that step that used to take 18 to 24 months of just aligning the data…… I’m not going to say ETL [extract, transform, and load] is dead, but it is less relevant in the agentic world.
HCI: UPMC Enterprises is one of your investors. That seems like an organization that would be an obvious potential customer.
Butt: We hope so. We cannot comment on that. They’ve been a fantastic partner for us. They were one of the early ones to to appreciate what it is that we were doing, and they’ve been supportive consistently throughout the process. I would say in our earliest days, the institution that recognized what we were doing was UPMC.
HCI: Is part of what interested investors the team you’ve been able to build in terms of execs with a lot of experience in healthcare?
Butt: Yes, three of the technical leaders — head of AI, head of delivery, and head of engineering — are ex-Optum people. Many of the team that we have either worked at Kaiser, United, Optum, CVS and others on the technical side. Then we have our chief strategy officer, Mark Caron, who was the CIO at three different Blues plans and a CTO at Catholic Health Initiatives, as well as the CIO of the Collaborative Care Division at Optum.
And Missy Krasner, who’s been an advisor for us from the beginning. She’s been at Amazon, Google Health, and Redesign Health. I think that shows in the completeness of the vision and and it shows in the understanding of the nuance on governance and security and privacy that is particular to healthcare.
HCI: Are there some things on the policy and regulatory side, or involving marketplace pressures, that are giving you tailwinds to get more engagement?
Butt: I think there are two big tailwinds. One is that payers have been decimated over the course of the last 18 months. The Medicare Advantage business that used to be highly profitable has turned and their medical loss ratios have gone through the roof. So there’s massive pressure on them to reduce costs. And the cost that they can control is not the MLR, it’s the operating cost of running the business.
Then CMS has some regulations around interoperability and specifically around prior authorization that people are trying to recalibrate around. With an agentic world becoming reality, a lot of people are choosing agents to do that automation vs. another application that they used to do. So we are seeing a tremendous amount of demand and a number of engagements on that use case on the payer side. Risk adjustment is another use case on the payer side that is quite robust for us. Claims is developing as another area of of demand. On the provider side, I would say it’s around revenue cycle management — the claims scrubbing, the extraction of medical codes. Medical coding itself is a big use case for us. And then appeals is the other one. We’re going to be launching some additional capabilities that we’re piloting around patients and providers that will be coming down soon.
HCI: Do some agentic efficiency gains to health systems or payers come from reducing administrative employee head counts? Or are they seen as tools for those employees to use?
Butt: I don’t see any replacement of head count anytime soon, frankly, and it’s because the system is capacity-constrained right now. The nurses or doctors or admin staffers are being asked to do more and more. An example is prior authorizations. One of the big payers told us their pharmacists used to have to review 40 cases a day for prior authorization. Then that went up to 70 cases a day. Now they’re asking their pharmacists to do 100-plus case reviews a day. And guess what’s happening? They’re burning out. They’re making mistakes. They’re quitting and leaving the field. I have a personal story around this, because my wife used to be a prior authorization pharmacist, and I saw her go from working eight hours a day to 14 hours a day trying to make quota. So I think it’s a capacity issue. It’ll just relieve the people who have the responsibility to do this work to focus on the most important part of the work and work at the top of their license.
David Raths is a Contributing Senior Editor for Healthcare Innovation, focusing on clinical informatics, learning health systems and value-based care transformation. He has been interviewing health system CIOs and CMIOs since 2006.
Follow him on Twitter @DavidRaths
source
This article was autogenerated from a news feed from CDO TIMES selected high quality news and research sources. There was no editorial review conducted beyond that by CDO TIMES staff. Need help with any of the topics in our articles? Schedule your free CDO TIMES Tech Navigator call today to stay ahead of the curve and gain insider advantages to propel your business!
