00:00.97 Ratnadeep Bhattacharjee Hello, and welcome to Leaders Perspective, where we sit down with healthcare and technology leaders redefining how innovation meets in impact. I'm your host, Ratnadi Bhattacharji, co-founder of Tech Variable. 00:13.21 Ratnadeep Bhattacharjee Today's episode is titled Smart Claims, Smarter Care. And we are ah diving deep into the critical intersection of claims automation, you know legacy modernization, and the emerging AI-driven transformation in healthcare. 00:28.69 Ratnadeep Bhattacharjee To help us unpack these, I am thrilled to be joined by Nate Perry Wistissel. you know I hope I pronounced it correctly, Nate. Nate is a seasoned ah technology and product executive with more than 20 years of experience scaling B2B SaaS platforms in complex regulated industries like healthcare. 00:49.11 Ratnadeep Bhattacharjee He has had large teams through cloud transformation, you know ai adoption, legacy modernization, He currently serves as the principal architect of the PTP engine, a proprietary agentic framework to accelerate strategic strategic execution through automation, orchestration, and intelligence. 01:06.29 Ratnadeep Bhattacharjee Nate, welcome to Leaders Perspective. It's a pleasure to have you here. 01:10.60 Nate Perry-Thistle Thank you very much. but deepep It's neat. Perry Thistle. 01:15.30 Ratnadeep Bhattacharjee Thank you. Thank you for the correction, Nate. ah Nate, I was just going through your article that you shared, you know, I guess you shared it on LinkedIn yesterday, right? 01:27.17 Ratnadeep Bhattacharjee The AI civil war in healthcare is here. And it really kind of struck a chord considering the kind of work that we at Tech Variable are also doing, right? ah You framed healthcare's evolution as a battle between legacy systems and ai native platforms, right? 01:43.60 Ratnadeep Bhattacharjee Especially in areas like billing and claims, right? And now, why do you see the claims process and billing, right? and building, you know, more broadly speaking as such a central battlefield in the AI civil war. 01:58.08 Ratnadeep Bhattacharjee What inspired you to speak about this now? 02:01.80 Nate Perry-Thistle Yeah, i actually was inspired to do some research by an article I read in Civil Beat here in ah Honolulu, which was really about how AI is making decisions about our healthcare, and we don't even know it ah The kinds of decisions that are being made are really ah kind of that old model, deterministic AI. Some may just be algorithms, but they essentially, ah you can put them into kind of the robo denials box. You know, they're a black box. They do something. 02:31.07 Nate Perry-Thistle And quite frequently, their motivation is to save money by preventing or not authorizing care. And it's creating a situation where things like the improvement standard, which is illegal in the United States, 02:44.74 Nate Perry-Thistle is creeping back into the system ah where claims are being denied because people are are needing that care simply to maintain or sustain, and and they aren't, in fact, going to get better. On the other side, you have the the new models of ai more nimble things, and I think there's this emerging battleground where there's an opportunity for AI to be used for human potential and human good and can really offset that black box robo-denials situation. 03:12.83 Nate Perry-Thistle So that's just an example where maybe there's a ah battle to be had where AI will be warring with itself. 03:27.08 Nate Perry-Thistle Oh, I'm sorry. I've lost audio. 03:30.37 Ratnadeep Bhattacharjee Yeah, yeah. Sorry. I mean, Nate, I understand that, you know, ah AI at the ah being at the forefront of every discussion within the closed doors of enterprises, right? 03:42.55 Ratnadeep Bhattacharjee But why do you think it's a war, right? Because as, I mean, if you scroll your LinkedIn, if you scroll down your social media, you will only find these decision makers or CXOs only talking about AI, agentic AI, gen AI and whatnot, right? 03:59.56 Ratnadeep Bhattacharjee Why do you think it's a war then? 04:02.70 Nate Perry-Thistle ah You know, the Civil War is ah is a great way to grab some attention and get people thinking about the opposing sides of ah the good or bad that we can do with AI. 04:13.45 Nate Perry-Thistle I think we all look at an industry like the 04:20.88 Nate Perry-Thistle computer ah security infrastructure, right? Heavily influenced by ai against AI endeavors right now, really head-to-head warfare ah going at our security infrastructure and our systems. 04:35.25 Nate Perry-Thistle The old things that used to work, like looking for poorly formatted emails or and yeah URLs, aren't going trick AI anymore. 04:40.64 Ratnadeep Bhattacharjee Thank 04:43.04 Nate Perry-Thistle So you have more sophisticated attacks and more sophisticated defenses. And I think that's just a kind of a microcosm model of how a lot of business will evolve. um 04:53.17 Ratnadeep Bhattacharjee you. 04:53.90 Nate Perry-Thistle I think you know there are examples in the healthcare care industry, for example, where ah you have value-based care creating a tailwind where doing the right thing for people, getting them cared for, being proactive about that care, and using AI for good creates the same kind of opportunities and benefits and cost savings that you would look for with that kind of poor application or bad application of AI where you're just using it to deny at scale. 05:20.60 Ratnadeep Bhattacharjee Thank you. 05:23.63 Nate Perry-Thistle I think one of the startling statistics that I read as I did the research was ah something like 90% of the claims being processed by this black box process were being denied. 05:36.24 Nate Perry-Thistle um When I'm sorry, I'm going to have to rewind that. But my fact is incorrect. um There are Tens of thousands of claims being denied by the system at high scale. 05:50.81 Nate Perry-Thistle ah When the systems are challenged, 90% of the time the claim is overturned. 05:52.40 Ratnadeep Bhattacharjee Thank 05:56.26 Nate Perry-Thistle The patient does, the patient does in fact need that care. 05:57.06 Ratnadeep Bhattacharjee you. 06:01.28 Nate Perry-Thistle Unfortunately, only 0.2% of the time are people actually asking for that appeal. So that's that kind of, again, good and bad opportunity I see in AI or that civil war that might be playing out in AI. 06:12.95 Nate Perry-Thistle There's a huge multi-billion dollar opportunity for somebody to do good for people and create the right kind of AI that is more transparent about things. 06:24.84 Ratnadeep Bhattacharjee yeah I mean, you touched upon a very ah important pointer regarding claims processing, right? You know, claims rejections nowadays are more than just an operational hiccup, as you can understand. As you as you rightly point out, they cost the industry billions, you know, they frustrate patients and in fact, even delay care, right? 06:46.15 Nate Perry-Thistle Yeah. 06:46.04 Ratnadeep Bhattacharjee What do you see the most you know preventable causes of rejections and how can smarter, probably agent tech preempt this before they happen? 06:58.30 Nate Perry-Thistle I think there's an opportunity to really treat the claims data like an event stream as we would in software engineering and use that to build smarter systems that are more capable of taking care of business based on things that are happening in real time. 07:16.43 Nate Perry-Thistle One example might be a system that actually sees parts or pieces of a claim are missing in real time and enables an auto correction to go back and get that piece so that the submission works. 07:29.87 Nate Perry-Thistle It's a very simple feedback loop, but just that little bit of error error correction in the system could create a better outcome in the future. I think if you can apply that at scale to lots of problems and look for opportunities and patterns that AI can solve that way, you really can treat that event stream as really an opportunity to do all kinds of good things in the moment to take better care of people, ah to produce better claims, ah to get paid and to you know earn a better line ah bottom line. 08:02.87 Ratnadeep Bhattacharjee Interesting. ah you know we Pre-COVID, at Tech Variable, we were working on something very similar. you know of claims We were developing a claims rejection ML model for a big health tech company. right you know it's It's a huge health tech company. 08:21.40 Ratnadeep Bhattacharjee So it took us about you know six to nine months to really understand the overall workflow of how things work and all the parameters involved and whatnot. right So I personally see a lot of you know ah use cases where I mean use cases of agentic AI as as as a framework to really make an impact in the entire ecosystem of RCM you know as a whole. I mean, it's a very, very manually driven process at this point of time. But yeah, I completely get it. 08:54.83 Ratnadeep Bhattacharjee with one of the ideas that you just pointed out, right? And one of the things that we previously worked on, I see a lot of things that can be done. Still, you know, I know people are doing it, but, you know, there's there's so many areas still untapped, I believe, right? 09:07.41 Ratnadeep Bhattacharjee You know, ah you've got to understand that too often automation is isolated from the bigger picture, right? According to you, Nate, how can intelligent claims processing drive broader business outcomes, like population health insights, network performance, and value-based care enablement? 09:29.89 Nate Perry-Thistle I think if I could zoom out, one of the things that I'm really excited about that I see coming is this key change in how we are applying AI. 09:40.95 Nate Perry-Thistle We're moving from these kind of monolithic systems, ah kind of back again to this Civil War theme, ah that are owned by the big guys into these more compositional systems where it isn't really what hardware you choose or what cloud provider you choose. 09:53.83 Ratnadeep Bhattacharjee Thank 09:56.79 Nate Perry-Thistle There are reasons to choose one or the other for certain jobs. But really, I think where the battleground is headed is the orchestration layer above that. It's not choosing one foundational model and large language model to solve a task. It's about choosing the right large language model to get the job done at the right price. 10:15.91 Nate Perry-Thistle And I think one of the things that you kind of mentioned as we move toward value-based care, I see this agentic workflow that we talked about, that event stream of the claims data, enabling all kinds of opportunities for more discreet, I see we've run out of space. Are we okay to keep going? 10:34.62 Ratnadeep Bhattacharjee please please. 10:35.79 Nate Perry-Thistle Okay, ah sorry. um So I see this opportunity where you have agents that can handle discrete jobs, and they can be really expert in them. And so I think there's a sea change coming for the enterprise too, where you no longer think about buying this large monolithic piece of software or the ERP system. 10:55.51 Nate Perry-Thistle You're really thinking about how can I buy the best of breed ah AI to solve a problem? How can I use an a agent from that to actually address a specific task or need? 11:06.51 Nate Perry-Thistle And I think we will see the kind of the startups and the monolithic, you know, the big players continue to duke it out in these ways again here still too. If I have a really abstracted data layer, good event stream, and I've got something like FHIR accessible, you know maybe wrapping my legacy data systems, I've really positioned myself well to be able to drop in any AI, the most affordable or the best to solve whatever problem I have in that moment. 11:18.45 Ratnadeep Bhattacharjee Thank you. 11:32.87 Nate Perry-Thistle so If it's overcoming a really sticky claims problem and somebody has the industry standard solution to that, as a health system, I want to be able to to so subscribe to that despite being you know having chosen where I might cloud host myself or co-locate myself. 11:48.37 Nate Perry-Thistle So I see this really compositional future coming. And I think event streams and treating the claims data that way is a great way to, kind of as a technologist, embrace this future, this agentic future. And lots of things will have to change in the enterprise. 12:02.31 Nate Perry-Thistle ah But we're all in a sea change these days. And I think we need to look forward and and look how we can work differently. 12:12.12 Ratnadeep Bhattacharjee Interesting. I think, Nate, what you are also trying to say is modernization of existing infrastructure would play a key role, right? But you've got to understand... 12:23.58 Ratnadeep Bhattacharjee Nate, I mean, of course you understand being an industry veteran, right? But modernization is hard, right? 12:29.20 Nate Perry-Thistle Yeah. 12:29.18 Ratnadeep Bhattacharjee Especially for payers and providers with deep investments in, you know, legacy infrastructure, so as to speak, right? What's your guidance for teams, you know, planning a phased migration to intelligent platforms while kind of minimizing, ah you know, disruption as much as possible? 12:48.57 Nate Perry-Thistle I think it's really important to invest in machine learning and making sure that you have good, clean data. ah you know First and foremost, you know how can you get to that kind of clean knowledge base, that good source of truth with your data? 13:04.68 Nate Perry-Thistle I think one of the interesting things is that Gen.AI is actually pretty good with messy data. And one of the opportunities we have here is to really just try to expose the data, to get it into a reasonable data lake and make it more accessible and more usable. 13:19.62 Nate Perry-Thistle I think one of the other opportunities that I alluded to is to try to abstract away your problems. There are wonderful technologies that can create data governance in a wrapper layer around your legacy system ah so that you can exchange data in a standard like FHIR adopt some of those modern protocols. 13:36.75 Ratnadeep Bhattacharjee Thank you. 13:41.94 Nate Perry-Thistle If you can do that and modernize your interactions, you open up a world of APIs. you open up the ah ability to drop in point solutions anywhere along those problems. And if you embrace that event-driven system, you can recognize when there's something in the system that needs either human attention or agentic attention, and you can get a more auto-correcting system in the mix. 14:07.82 Ratnadeep Bhattacharjee Interesting. I mean, one of the things or rather one of the solutions that we have internally, which kind of has played very important role in within a few of the health systems has been our solution SyncMesh, right? It's an it's an interop layer that we have, which is the first step, you know, you you should ensure interoperability, you should have all your system systems, you know, aggregated, integrated properly, you know, so that you can actually create those data warehouses and data lake to create your own use cases on top of those, you know, master data layer or whatever you will call it, right, golden records. 14:45.13 Ratnadeep Bhattacharjee So I completely get it, right? That's how you plan an intelligent kind of migration and phased migration, you know, slowly but surely. You have to understand what your use cases are before everything else. 14:59.48 Nate Perry-Thistle Yeah, I think you're hitting on, you know, you have to solve a real problem first and foremost. 14:59.56 Ratnadeep Bhattacharjee So, 15:04.08 Nate Perry-Thistle i And, you know, unlocking your data and making it accessible is is job one and being able to get that done. 15:11.88 Ratnadeep Bhattacharjee So in your, um you know, huge 20 years experience, right? Have you kind of faced issues where, ah wherein you have approached or you have been part of an organization? 15:27.70 Ratnadeep Bhattacharjee wherein they want to do something, but they've not been able to for certain reasons, you know, because it may be because of some technology debt or, you know, probably legacy infrastructure seeped into their entire organization. 15:41.86 Ratnadeep Bhattacharjee Have you faced anything of that sort, Nate, ever? 15:46.28 Nate Perry-Thistle Yeah, I mean, I think the world, you know, sort of for me has boiled into a couple of different ah approaches. I get to start something fresh and I am the monster and I create the the you know the whole architecture. And ah sometimes that's a lot of fun. We get it right. ah You know, it's modern, it's nimble, it's mobile first, it's ah've you know got a great CI CD stack. Maybe you've got a mono repo in play and, you know, just everything is hunky dory. 16:16.33 Nate Perry-Thistle more likely you've got something really valuable in a monolith or an old piece of software that does a venerable job and it's really hard to get away from it. 16:22.58 Ratnadeep Bhattacharjee you. 16:25.79 Nate Perry-Thistle So I have practiced a lot of the Strangler Fig pattern. I'm a big fan of playbooks and approaches that allow you to come back and you know bring a repeatable approach to things. 16:37.40 Nate Perry-Thistle And so you know one of the things that I've embraced is the idea that you need to really be data first. 16:38.85 Ratnadeep Bhattacharjee Thank you. 16:43.85 Nate Perry-Thistle right I've talked about this a little bit. So when you find yourself you know kind of mired in the monolith, I think one of the things you can do is is take that data first approach. How do i transform from an ETL mentality into an ELT mentality, where I've got lots of data flowing into a lake or many puddles, ponds, and lakes, um and really you know kind of get that mentality. 17:08.62 Nate Perry-Thistle Get into the event-driven architecture. You have an event log in your database, in your monolith. It is already spitting out events that you could hook a modern system into. treat your monolith as an API to be abstracted away. 17:22.03 Nate Perry-Thistle ah Figure out what most what its most important functions and features are, and then create an abstraction layer around that. And I'll vote again for solutions like GraphQL that let you create a wrapper around things that abstract kind of that front-end presentation and getting and setting of information 17:37.89 Ratnadeep Bhattacharjee Thank 17:41.54 Nate Perry-Thistle away from the heavier lifting of defining the data models and abstracting away complex APIs. And as you know, with GraphQL, you could be interacting with all kinds of things. 17:52.23 Nate Perry-Thistle I think a couple of other points I would add on in this modern world, I really think you'd need to be thinking about how to be multi-cloud capable. ah Each cloud has its ah rationale and reason for being there, but I think you know understanding when to use each is really important. 18:09.29 Nate Perry-Thistle And I also think as we embrace ai it's not just about picking the winning model anymore. 18:10.84 Ratnadeep Bhattacharjee Thank you. 18:15.60 Nate Perry-Thistle It's about abstracting yourself from those models with something like Vercel or Langchain or something that allows you to pick and choose which foundational model and which large language model to use and what version so that you can control your costs. 18:31.66 Nate Perry-Thistle And I think that that abstraction will serve you well as you think about that kind of data first mentality and setting yourself up for a future where we really don't know what's coming in the next few years. It's about agility. It's about playbooks. It's about finding ways to orchestrate that and create those repeatable processes. 18:50.91 Ratnadeep Bhattacharjee Interesting. Now to a completely, you know, not separate topic, but you know, something which I am intrigued about, you know, you're building something ah truly unique with the PTP engine, right? Can you, can you walk us through what an agentic AI framework means in the context of healthcare claims? 19:10.94 Ratnadeep Bhattacharjee You know, how does this differ from traditional, you know, RPA or rule-based systems? What are some and probably early results or breakthroughs you have seen and with this approach in the real world setting settings right 19:24.97 Nate Perry-Thistle Yeah, well, I'll confess that my work here is theoretical still. I do have what I lovingly call the Perry Thistle Playbook Engine. And it really is the idea that we as executives need to shift right in our culture a little bit. 19:39.05 Nate Perry-Thistle I think we're asking our product managers and our software engineers to shift left a little bit. and we're all coming to the center a little bit. So as an executive, you know this is really me putting hands on keyboard and having an opinion about where the future is headed. 19:52.39 Nate Perry-Thistle So i've I've teased a few of those things that I think are really fundamental. fundamental um you know Knowledge is everything. ah Context is key, right, in the agentic world. We really need to ground our agents in great context. And that's sort of job one, the most important thing for you as the executive. 20:09.30 Nate Perry-Thistle unlock your data, spend the money now, solve the problem, whatever it is, abstract it away ah because you need that agility. And then the other piece I mentioned is playbooks. 20:19.73 Nate Perry-Thistle Really our work as executives is about figuring out how to optimize organizations to go and take objectives over and over and over again. 20:24.26 Ratnadeep Bhattacharjee you. 20:28.34 Nate Perry-Thistle We've really learned to love OKRs and smart goals. Maybe it's time to learn some different goal structures that are more suited to this compositional future that I'm alluding to. 20:38.92 Nate Perry-Thistle And so what I've been working on is the idea of a compositional AI system that's choosing best of breed solutions to help me solve problems. The ontology is fairly simple. 20:50.92 Nate Perry-Thistle I have objectives, currently three, that I need to meet. ah Each objective, I have a strategy ah laid out to guide how we go about meeting that. 21:01.77 Nate Perry-Thistle objective. And then we apply playbooks. 21:03.54 Ratnadeep Bhattacharjee Thank 21:04.37 Nate Perry-Thistle And playbooks are really where I have agents get involved in the the efforts. 21:08.89 Ratnadeep Bhattacharjee you. 21:10.16 Nate Perry-Thistle The agents take the form of several tiers. My kind of top tier agents are orchestrators, something I call the chief of staff. And I apply a mirror, augment, and complement framework to create something that thinks like me and complements the way I think and challenges the way I think. 21:21.33 Ratnadeep Bhattacharjee Thank 21:24.81 Ratnadeep Bhattacharjee you. 21:26.27 Nate Perry-Thistle I have a number of tier two ah agents that are capable of specific tasks. I think, ah example for an example, a historian is a really easy one. it is able to create knowledge, ah for example, to feed a RAG ah system so that you have that retrieval augmented generation system, you know, fresh with knowledge data. 21:47.28 Nate Perry-Thistle And I also have Forge who is able to write software. ah Forge has actually been imbued with some power, so he can run auto gen crew, ah for example. So we have orchestrators and we have kind of these tier two agents that it can actually do things. 21:56.62 Ratnadeep Bhattacharjee Thank you. 22:00.34 Nate Perry-Thistle Tier two agents, think of those as they can interact with the model context protocol. They can go and interact with the real world on your behalf. ah There's a client server protocol in MCP, so they can go and do things as simple as something in GitHub or submitting a claim for you, perhaps, through some system that has an API for it. 22:20.91 Nate Perry-Thistle um And then finally, I have tier three agents, which I think are very composable. This is where the secret is in kind of the prompts. ah Prompting works the best when you give it great examples, you ground it in knowledge. 22:30.21 Ratnadeep Bhattacharjee Thank you. 22:33.46 Nate Perry-Thistle And that's really what my my system is geared toward doing. So we have objectives, we have the strategies to achieve them, we have playbooks, and then we send our our agents out on missions to orchestrate and achieve these little goals along the way. If you think about it, 22:48.94 Nate Perry-Thistle It's very much like organizing a team at work. You are the mini-CEO of your agentic empire, and it's out there doing work on your behalf. And I've had some really amazing results. 23:01.08 Nate Perry-Thistle So where I think this is headed is that a healthcare care system, for example, might no longer subscribe to a single large monolithic piece of software, but they might buy a best-in-breed agentic solution to solve a particular part of that chain. 23:16.42 Nate Perry-Thistle And I think this creates some opportunities for new winners in this new economy, this agent-to-agent compositional future that we're headed for. um The people who are creating the agents, including the best agents to solve something like a claims problem or to combat that robo-denials problem, I think those folks are in a position to really be clear winners. 23:27.73 Ratnadeep Bhattacharjee Thank you. 23:41.91 Nate Perry-Thistle Another great group of people, and partly why I'm working on this thesis myself, is that our future is compositional and orchestrated. And so orchestration is going to become this new competence that we all must embrace as executives ah to be able to excel in the new world of work. 24:01.30 Ratnadeep Bhattacharjee Interesting. So, Nate, how do you create the enterprise lift, right? You know, it's about orchestrating outcomes rather than just automating steps, right? How do you do that? 24:15.51 Ratnadeep Bhattacharjee Okay. 24:16.66 Nate Perry-Thistle I think it comes back to one of those fundamental questions, the why of things, right? ah So how do you get at that organizational 24:28.66 Nate Perry-Thistle impetus, the need for change, that change management, those disciplines will still just be a huge part of what we do. um I think we need to really focus on the results. 24:39.52 Nate Perry-Thistle What is it that we want to get out of something? Are we achieving them? We're going to have this ability to do so much so quickly in the coming years with AI. we're really multiplying our own power. 24:52.61 Nate Perry-Thistle Are we setting ourselves up with the organizational muscle to measure those outputs in a new way? A different return on investment, for example, for AI-driven workflows can't just be about saving money or saving time. 25:07.90 Nate Perry-Thistle There's all kinds of new ways to quantify those experiences. So I think it falls down to good old-fashioned change management, rolling up your sleeves, being an inspirational leader, and really having a thesis and getting into that kind of you know modern, goal-driven approach. Get your team to become problem solvers. 25:27.07 Nate Perry-Thistle ah Give them a hypothesis to test. help them unlock the puzzle for you ah and challenge them to think that way. Think in quarters, don't think in years. 25:38.06 Nate Perry-Thistle um you know Do think in years, but solve problems quickly and and make sure that you're really focused on that core, fundamental piece of work that you need to do. 25:50.30 Ratnadeep Bhattacharjee Interesting, you know, baby steps. I get it. 25:53.43 Nate Perry-Thistle Yeah. 25:53.71 Ratnadeep Bhattacharjee So, Nate, how, and I mean, you have kind of conceptualized this PTP framework, right? You know, what really, especially within healthcare settings, right? 26:04.56 Ratnadeep Bhattacharjee What will make this framework or platform, you know, governed and resilient, if I may, if you know what I mean, right? 26:14.07 Nate Perry-Thistle Yeah, I think maybe coming back to our opening theme and the idea of that um little bit of an AI civil war and the sort of the closed systems versus the open systems again. 26:18.85 Ratnadeep Bhattacharjee Thank you. 26:27.30 Nate Perry-Thistle And so I think if you look at a ah compositional AI future where we might have had a black box in the past that some part of the process was denied or a claim wasn't handled, we'll have a little bit more transparency required. 26:42.32 Nate Perry-Thistle um One of the reasons for that will be that we can't really regulate things at the foundational model level anymore. like The way that changes, it's just too quick. 26:53.12 Nate Perry-Thistle ah But I think what you can regulate is at that orchestration level. Do you have a system that can transparently describe all of the decisions that were made in the system when a human in the loop intervened in that decision-making process um and create that really transparent system that allows people to understand things? If you have an audible end-to-end system, ah then I think you can really earn trust and trust. 27:24.38 Nate Perry-Thistle Yeah, so I lost the thought there. I'll have to edit that one out, please. 27:27.78 Ratnadeep Bhattacharjee That's fine. That's fine. 27:28.56 Nate Perry-Thistle Okay. 27:28.38 Ratnadeep Bhattacharjee I get what you're trying to say, Nate. I get it, right? Nate, you know, a good segue to this would be, you know, I wanted to also understand, you know, having spoken with you a couple of times and, you know, having read through most of the stuff that you kind of opinionated on, you've been a strong advocate for designing AI native products, right? 27:51.05 Ratnadeep Bhattacharjee You know, instead of just leveraging AI onto old systems. What are the core design principles that guide you when building in regulated spaces? And how do you balance automation and trust, especially, you know, as you said, human in the loop, right? Especially where human oversight is essential. 28:09.58 Nate Perry-Thistle Yeah. um Well, I think one of the things that you're alluding to is we really have to earn trust with human beings in these systems. So kind of going back to that transparency and the explainability of the system, I think that's really important. So one of the things I think is how we frame the tools that we create. 28:29.01 Nate Perry-Thistle um ah and how they can be used and how we think about the user experience of those tools. For example, if you present something to a clinician and say, this user has a readmission risk, 28:45.26 Nate Perry-Thistle they're probably not going to be very enthusiastic about working with that from an AI model. But if you can say this person has an increased readmission risk because of these blood pressure results that we see increasing over time and that emergency visit and this other piece of information, 29:04.47 Nate Perry-Thistle you know Then I think what you're creating is a co-pilot where the clinician feels like, okay, I have something that's helping me. It's making connections for me. It's helping me see things. It might not even be making recommendations at the start, right? 29:18.29 Nate Perry-Thistle It's just making connections. And I think that's one of those things that we have to help people see is that... It's really just a tool. It is the human that makes these tools valuable. 29:27.81 Ratnadeep Bhattacharjee you. 29:28.21 Nate Perry-Thistle And so we want to think about building ai for humans in a way that empowers them, that presents what they're using as a complement to them and not as a replacement for them. 29:42.65 Ratnadeep Bhattacharjee this is This is what I was you know trying to understand when you when you kind of alluded somewhere. you know I forgot where you mentioned it, but AI native products. right So I get it now. 29:56.33 Ratnadeep Bhattacharjee Nate, looking ahead, right what's your vision for where claims processes are heading now? Are we close to seeing real-time context-aware or self-optimizing claim systems? 30:09.83 Ratnadeep Bhattacharjee how do you i mean how can this reshape both payer and provider strategy you know does where do you see all this going you know moving forward 30:20.61 Nate Perry-Thistle Wow. I think that we are going to be in for a turbulent few months and few years. We have um systems that are changing at a pace that can't be regulated and can barely be kept up with. 30:37.10 Nate Perry-Thistle On the other hand, you have a need to, you know, kind of keep up with those changes and solve problems for people. We, I continue to see this need to transition to value-based care as a catalyst ah to motivate us to, you know, kind of adopt these models that are better for humans. 30:57.66 Nate Perry-Thistle um 31:03.22 Nate Perry-Thistle I think that 31:06.57 Nate Perry-Thistle There are solutions already where AI is kind of creeping in, and um we're going to see a lot of lawsuits probably in the you know coming ah months or years play out in this space. 31:19.10 Nate Perry-Thistle So I still am optimistic that there is an opportunity here for the right kind of AI to really improve the visibility into the claims process ah to make ah you know better predictions about what will ensure your claim gets paid, to improve the interconnectivity of our systems so that it is easier to do things. 31:30.94 Ratnadeep Bhattacharjee Thank 31:39.35 Ratnadeep Bhattacharjee you. 31:42.29 Nate Perry-Thistle And again, like it it it needs this internalized ah energy from the industry to do good for people. right It can't be all in service of ah profits from and you know purely insurance perspective. You have to embrace that value-based care perspective. And by taking care of people, I think you know testing that hypothesis, again, I'll bring up a Kaiser model, you can create a system that is motivated to take care of people and in doing so ensures profits for the system. 32:13.51 Ratnadeep Bhattacharjee that's that's so what most importance it has to ultimately give an enterprise outcome or else nothing will make sense right so uh net for healthcare care leaders uh you know healthcare get tech leaders listening in to this podcast right what would you say is the first mind shift 32:20.67 Nate Perry-Thistle Yeah. 32:31.46 Ratnadeep Bhattacharjee mind shift change uh they need to make in order to embrace agentic ai you know agentic and native thinking if i may say so right what's the what's the first step towards you know transforming especially claims right from a burden into a strategic level 32:51.93 Nate Perry-Thistle I'm sorry, will you pause just a moment? 32:56.35 Ratnadeep Bhattacharjee so nate uh did you get the question i asked like uh 33:01.77 Nate Perry-Thistle No, if you could respond, please. 33:01.93 Ratnadeep Bhattacharjee i was ah yeah as I was asking that, you know, for healthcare tech leaders, right, you know, who are listening into the podcast, what would you say is the mind shift or mindset shift, you know, they need to make in order to embrace, you know, agentic AI native thinking, right? What's the first step basically towards transforming something like claims, which is something that we discussed today about, 33:29.08 Ratnadeep Bhattacharjee from a burden to kind of a strategic strategic leverage. 33:32.67 Nate Perry-Thistle Yeah. Well, I think I'm going to come back to the open or closed ecosystem choice. So figure out where you are in that. It should you know start with a dispassionate assessment of what you need. right So think about your organizational ah organization risk tolerance. Do you really need to be in that enterprise best in class, very stable system, or do you have the opportunity to move a little more quickly and and take advantage of other solutions. 34:02.61 Nate Perry-Thistle um So first and foremost, figure out if that open or closed ecosystem is for you. ah The next thing I would recommend is think about orchestration. How can you embody orchestration in your organization so that change management becomes a muscle? 34:18.00 Nate Perry-Thistle What are the playbooks that become your intellectual property that allow you to create value in your organization? Then I would say solve your data problems. Make sure that your data plumbing allows you to build in the future. Do the unglamorous work now. Make the investment. 34:36.14 Nate Perry-Thistle It is the foundation upon which all of your future rests. AI is voracious for data. Then I would recommend thinking about finding something small to get into, ah something that is meaningful and solves a problem, ah like burnout for clinicians. Do something that isn't challenging their work and their expertise, but get something out of their way, ah that 25% of their time that's wasted doing administrative tasks, for example. 35:08.44 Nate Perry-Thistle um I know that's kind of a lot of things, but I think they all go together. That's the mindset shift that you have to make into these things. 35:17.32 Ratnadeep Bhattacharjee Nate, this has been incredibly enlightening. You have helped us see, and you know, claims not just as a back office cost center, but, you know, kind of a strategic engine that can power better care, faster operations and smarter systems if used properly. Right. 35:33.91 Ratnadeep Bhattacharjee To our listeners, if you found this insightful, don't forget to subscribe and share. These are the conversations that may drive real change. Nate, where can our audience follow our work, follow your work or learn from you about the BTP engine that you are kind of developing now? 35:51.64 Nate Perry-Thistle You can find me on LinkedIn as Nate Perry Thistle. you search the internet, I'm pretty much the only one. 35:59.15 Ratnadeep Bhattacharjee Okay. ah Thank you again, Nit. And thank you all for tuning in. This is Ratnadeep Pacharji and you have been listening to Leaders Perspective. See you in the next episode.
