The Workers’ Compensation sector has a rich history of financial success in underwriting, but the landscape is evolving rapidly. The need to enhance operational efficiency, combat fraud, and navigate the intensifying competition from multi-line insurers is more pressing than ever. Explore how Artificial Intelligence (AI), is reshaping the workers’ compensation landscape. Learn how insurers use AI to gain insights, streamline processes, automate decisions, and improve outcomes for injured workers. Presented by Clara Analytics and Origami Risk, this discussion reveals how organizations are advancing Workers’ Compensation programs by: Strengthening communication, streamlining care coordination, and accelerating return-to-work strategies to improve worker satisfaction and reduce claim durations. Leveraging cloud-based core insurance solutions to increase operational efficiency and support scalable growth. Applying AI in real-world workers’ comp scenarios to enhance insurer outcomes and drive profitability. Speakers: Jaime Henry, VP of Product, Origami Risk Heather Wilson, CEO CLARA Analytics Hello, and welcome to today’s Business Insurance webinar presented by Origami Risk. Today in our in the next sixty minutes, we’re gonna have an, engaging discussion on the use of artificial intelligence in, workers’ compensation. And, I I’m Regis Kosha representing business insurance, and I’ll be today’s moderator. I’ll be discussing this topic with Jamie Henry of Origami Risk and Heather Wilson of Clara Analytics. Delighted to have both of you here with us. I wanted to let you know that this is an interactive webinar, and we’re gonna devote some time at the end to the audience’s questions. So if you would, please use the the tool on your on your screens to send questions to the panelists, and, we’re gonna try to have time for all of those. But, certainly, we, we invite you to ask questions, during during and throughout, and we’ll we’ll devote some time for q and a at the end. Again, thank you, and appreciate you joining us for today’s discussion. So, Jamie, you are the vice president of product at Origami Risk and someone who has spent a lot of time looking at how technology can improve things in the insurance industry. And, Heather, with ClaroAnalytics, you have a a special relationship with artificial intelligence and emerging technologies that are making a difference. Tell the audience a little bit about your each of your roles and kinda what you’ve been focusing on relative to the topic at hand in workers’ comp. Jamie, would you like to Happy to. So I’ve been with Origami for over eight years now, and the majority of my career has been focused in risk and insurance software. I spent the first part of my career really working closely with clients on implementations, ongoing support and account management. And so that really allowed me to be able to see firsthand the challenges that our clients were seeing, certainly over a variety of coverages and lines of business, but certainly a lot of time on the workers’ comp side. So I now, as the VP of products for Origami, have the pleasure of driving our product strategy for our entire organization in the two divisions that we serve and a wide portfolio of clients. We just hit one thousand clients last week, which is a really exciting milestone. And we really pride ourselves on being not just a vendor, but really a partner to our clients. And we’ve done that and especially through a long history of workers’ comp specifically. So we’ll we’ll be diving in, obviously, on the topic of AI. So I’m really, really excited about the conversation we’re gonna have here together. Sure. I’ll take it after, after Jamie. So I’m Heather Wilson. I’m the CEO of Clara Analytics. But I have spent probably, the last two decades working in large enterprise organizations trying to build machine learning applications across several industries, but also within a carrier. So I have a lot of passion, around helping the adjuster and thinking through how do we help he or she direct those cases and really thinking about, you know, we are an AI claims organization. We have that platform. We our legacies, we started in workers’ compensation now in all casualty lines, but really that legacy of of that data and workers’ compensation. I think our passion is really thinking about, you know, making sure that human in the loop, that adjuster has a tool next to them and working in partnership with Origami, and we’ll talk about this a little bit later. But how do we always help the adjuster surface up things that maybe the human eye wouldn’t see? We’ll talk about this a little bit more, but Claire is absolutely trying to think about lost costs and thinking about the injured worker and directing the cases. And if you work with a TPA, how do we give you more visibility and oversight into that relationship as well? So more to come as we talk through, but really happy to be here with Jamie. Great. Well, thank you both. So the the line of business at hand is workers’ comp. And as we know, that’s enjoyed a relatively long history of of being a profitable line for for workers’ comp insurers. But there’s a lot of pressure still. There’s opportunity for operational efficiency gains, reducing, you know, some some premium leakage there, and just kinda, you know, overall remaining competitive. It’s it’s a line that tends to be quite competitive. So I I’m curious kind of for both of your thoughts on where do you see the greatest opportunities in workers’ comp for insurers to keep pace with these changes? And and where where do you see, let’s say, you know, technologies, particularly artificial intelligence, making inroads in terms of the operational efficiency piece to be, you know, to be even even better at it than they already are? I think it’s helpful to kinda start with the baseline that sets up excellence and success when it comes to the being able to do this adoption around AI and that larger AI umbrella term and kind of everything that it means. And again, this conversation will help us kind of drive into those explanations, get some education and talking about some use cases and opportunities there. Regis, if you could go to the slide, I want to talk about a use case here in a case study we have with one of our clients, AEU. And this is actually a shared client between Origami and Clara. But I want to start with, again, that baseline. So AEU came to us in twenty nineteen and they were looking for, again, that partner, right? Not just a vendor, but someone who was able to help them in their claims administration move from a legacy system to a more modern technology that we pride ourselves on. And not only were they looking to kind of create that operational efficiency that you’re asking about, Regis, for their own organization, but also for the clients and the members that they serve. So they needed to make sure that, of course, they had that baseline functionality that you would expect in a claims administration system, but they also needed the capabilities that allowed them to create some custom workflows and the unique configurations necessary to support the specific operations of AEU. So we can run to that next slide just to talk about kind of some of the outputs and results that they got from their implementation. So the team was not only able to kind of move to what was a more efficient system of working through a claim, like the life cycle of a claim and all aspects of the communication and the financials related to that. But what was a really important milestone and now what becomes this foundation for them in their partnership with Clara and the way that that information feeds into Origami is having that insight into the data. So for an adjuster themselves, it’s about that, of course, efficiency and intuitiveness of working a claim, but also the own insight that they have in being able to see what they need to do that day, that week, and that month. And then of course, kind of rolling that up at the leadership level. So the leadership now has the ability to see where there’s opportunities, where they need to dive in and make some adjustments at a very basic kind of, not basic, it’s an improvement certainly from what they’ve been experiencing previously, but basic in comparison to the way that you can layer, start to layer AI on top of that. So where might there be problems? Being able to see the claims holistically in a way they hadn’t before. So getting back to that question about how do we improve operational efficiency, sometimes it’s with just ensuring that you’re getting focused on the foundations of claims administration and looking for even the tiniest opportunities, because those tiny opportunities where you can reduce a click, where you can put some sort of autonomous workflow efficiency in place, those become cumulatively incredibly impactful over time. So if I can just jump off to piggyback on to what Jamie was saying. So we love our partnership with Origami. It gives such a great foundation when we are able to work together. What happens next is if you think about in your own personal life, if you use a GPS, some type of system, you know, you’ve got the human and the lip. You’ve got the driver just like the adjuster is driving his or her cases. You have done these routes many times. Okay? You get in the car, you’ve driven that route from work to home or home to the grocery store or whatever, you know, those ten miles or so. But what you don’t know and your GPS system knows this about you, knows your behaviors, it’s learned. That’s what machine learning does. We know the behaviors of the cases. We baseline them. But what you don’t know is you get in there in your mindset. Like I know how to go from A to B and direct the case. But what happens when during the day construction is in there or there’s, you know, a different zone for traffic, you know, speed zone, or there’s an accident ahead. These are the things that the GPS system will pick up when you sit down to drive from A to B that you’ve done every day because you have the experience. And this is how we think about AI with our workers’ comp claims platform. So what we’re trying to do is give you a second set of eyes, you know, and we’ve already had this great platform, Origami, that’s so easy for us to bring the data into an AI, Clara app. And we’re saying, okay, here’s the optimal path just like your GPS will tell you. Here is your optimal path at their first notice of loss because our machine models have learned over ten years all these different bodily injury cases, all the different ways from a cost perspective, from procedures, from providers in that particular geographic location that we would direct you to look at from a scoring engine. And all the time you have a second set of eyes looking over your cases. Okay? And wow, you know, think about that GPS or that air traffic control saying, listen, something happened. The claims data came through the night onto this case. You wake up in the morning, you log into the app, and it’s like something’s going on in the case. We’re really, you need to talk to the injured worker around this. Or speak to the TPA about what’s happening here. Maybe something happening, a procedure that looks off from what should be in the right care path and case path for that. Maybe there’s something with medication. Maybe there’s something with return to work. All of these things that we know experience adjusters know because they know the routes on these cases. But we also know that we are human. And while we keep the human in the loop, wow, I can have these AI models looking over these cases twenty four seven for me and go in to the case when the models are detecting something, surfacing something. We’re human. We know our eyes get tired. We know we have a they have a lot of cases, and we want the adjusters to be able to always have empathy for the claimants. And to give empathy to an adjuster is to give them a tool that works for them in looking through these cases. You know I love with the GPS analogy, and I think it’s really important to call out with some of the fears that can happen around AI, which is it, gonna take us over. Right? Like, AI is gonna take over our job. There was that recent study that came out of MIT saying that’s not gonna happen anytime soon. Agreed. Right? So we know that AI capabilities, just what Heather’s describing here, are supplements. They’re enablers to our human thinking, and there are sometimes those kind of corrections just like a GPS map, right? Even if you know where you’re going from A to B, being able to plug the directions in the GPS map, avoid that construction that Heather was talking about, avoid that traffic, right? Let’s give our brain just a momentary break so that we can use that for the more complicated decisions. And we all know the complications that come with administering workers’ comp claim. All of the regulatory requirements aligned with them, they’re complicated. So being able to have that AI sidekick, if you will, is critical. It is. We know that our clients in the insurance ecosystem is dealing with you know, some aging of the the population, retiring population of agents. And so how do we keep that, you know, knowledge somewhere, especially around workers’ compensation when that experience adjuster knows which providers in their network. Right? How do you translate that knowledge to the next generations that are coming in? How do you translate to them? These are the defense councils to really work towards and go against those, the claimant litigator. If it gets to that, that’s knowledge transfer. And so all of that you know, is based in an AI platform that is so much easier. I mean, many people are still remote. It’s hard to have the knowledge sharing like we used to before the pandemic, but you can do that through an AI platform and share that knowledge and know in our ecosystem, these are the physicians that are the best ones, the providers dealing with them. And if it happens that the score is not, you know, what you thought it was, guess what? You have other panels to look at and and and other options. So I think there’s that that’s also happening in that adjuster population. I think the second thing that people really maybe hold them back, like Jamie was saying, other than, oh, this AI is gonna take over. Absolutely not. Right? Just like the pilot, the human in the loop, just like the driver, we need the human in the loop. I think secondly, everybody thinks their data has to be perfect in order to jump into this AI, you know, have an AI playbook and get into, you know, working with an AI platform. And that’s the beauty too when you have, you know, partnerships like Clara and Origami. And also, you know, we also if you don’t have the data in the right way that Origami does, you know, expose and help us load into our tools, we also that’s the beauty of working with an InsurTech too. We’re not a carrier, but we’re there to do that engineering of your data where whatever form it’s in, wherever it is, to map it and to cleanse it and to transform it so that it can be ready for AI machine learning models. I think a lot of folks think my data has to be in this certain form or, you know, be in a certain place. And we do so much of that work. It’s easier when you’re like I said, when you’re working with a platform for origami. But I think some of that also needs to, you know, also be, you know, not not a myth anymore. Like, really, you you can work with these platforms in whatever form of your data it’s in. One of the things that in talking with the AEU team and some of the benefits that they’ve seen in their use of Origami, the partnership with Clara and speaking to that data cleanliness, is that it actually uncovers it in a more efficient way than maybe some manual interventions. You certainly can put many of those in place. Let me put some sort of data validations in. Let me put some sort of automation in. But the fact is we’re still humans working things and working many things at the same time. So I think what’s fascinating about when you’re starting to get these alerts back from like as AU was with Clara, then it automatically starts to trigger those opportunities where you can improve data cleanliness and then proactively put processes in place to ensure it stays clean, to ensure that the user, the adjuster is knowing where to touch. So that starts to kind of weave a little bit back to your question as well, Regis, around that operational efficiency and how it all kind of comes into play. Yeah, and I would just add too, Most of our clients are seeing anywhere, you know, from a five x, even up to twenty x and using our AI platform. And where does that come from? So one, because we’ve been scoring providers for the last ten years, we know the optimal path for that claims case. So scoring those bodily injuries and how these physicians have been treating them, the procedures and obviously treatments and so forth, we’ve been able to score them from A through E. And most of our workers’ comp clients most of that ROI has come up from really taking out the c, d, and e score providers out of their network. And and that’s all done through objective data. You know, our models read all of the medical bills that are coming in from the medical bill reviewer of that perspective carrier self insured. And so we’re able then, as they start to optimize their network, it really translates into getting the right physicians in our network that then really have that lens around costs. But it also translates into doing the right thing for the injured worker too, because you want to get them back to a healthy place for their families and for the employer. And so, yes, it helps financially, right, in working for the carrier, but there’s also a place like you’re giving them, directing the right care there. I think that’s one area over and over again we just see, and it’s incremental because we understand when we’re onboarding a client, we ask for five years of history of data because we have to baseline. We have to understand, we, our models and our data science team have to understand how have you treated these cases so then we can understand attribution of our contribution to the incrementality of the loss costs and helping with that. I think that the other place too is we really try to take surprises out of these cases. So, you know, we do have a severity modeling, right? Triaging just like you triage, you know, an emergency room, right? So how do you prioritize here? How do you know the severity? But because the AI models never sleep, you know, if something becomes more severe, it’s not this, you know, surprise that you find out a couple of quarters later and then the reserve hasn’t been informed. You know, the models don’t sleep. So they’re telling you, giving you indication and prompting, this looks like it’s getting worse than better. And you might wanna take a look on the reserve side of it too. So again, you know, just like the GPS analogy, we’re trying to take the surprises and help and direct that route and do it in a way that is helping you because we know we can only make so many decisions in a day. So how do you prioritize those claims for that day based on that severity and then what’s going on with those cases? Yeah. That’s all really interesting. I I love the GPS analogy. I think that works on so many levels. And, of course, it’s relatable because so so many of us are accustomed to using that. I mean, I I know even though, you know, I’ve I’ve lived in the same community for almost thirteen years and I I know my way from, you know, one end of town to the other. It’s still I just say, when you plug that in, it gives you additional insight and additional data that you can use. I I like to think in some ways it makes me a better driver, certainly a more aware driver. You know, how that can be applied, I think, in in the workers’ comp claims context is is we all know that, you know, the the industry has, you know, a bit of a talent a talent overhang in terms of a lot of experience. And and, Heather, you had mentioned the knowledge transfer piece. I I’m curious for kinda how you see the impact of AI, you know, on kind of the next generation of workers’ comp adjusters and kind of bringing people, you know, into into that industry who are replacing those who are retiring and, you know, kind of recapturing some of the knowledge and experience that that they have. You know, GPS is certainly providing, I mean, that’s a great example for providing, hey, here’s what’s going on or this has just come up you should know about on the route to a claim. But how do you see that as kind of preparing the next generation of claims professionals in workers’ comp? Right. So like you said, not only is it GPS, and we love when, experienced adjusters, whenever there’s an alert that is served up to them or surfaced or information that many times they’ll be like, I knew that. I’m like, that’s great. That means our models are working. And and then our models keep learning because of of that human and the loop too. And so especially if they take the alert and if they don’t take the alert, our models still learn because there’s a reason why they don’t wanna take the alert. So I think you also have to think about the GPS being kind of a knowledge repository as well for the next generations coming in. So being able for, you know, remote, think about an experienced adjuster, being able to bring up on a second or third screen. Right? These board adjusters have all these screens that they’re dealing with. You know, having their Zoom, having their core system in the middle, origami, and then having their AI signal on the other side. So they can sit there on Zoom on their first and help a an adjuster maybe coming in from, you know, the next generation explain to them, here’s how I’m dealing with this super complex workers’ comp case. This is how I’m doing the administration piece and the insights within Origami. And then here I am within my AI signal application and where I’m getting direction. And let me show you all the pieces. Here’s the claim summarization, which is like the cliff notes where the claim is in the life cycle. Here’s the explanatory why these alerts are coming up. It’s almost like you can do a train the trainer because you have the tools. It’s not all these records, paper records. And I think it becomes a lot easier to transfer that knowledge to this next generation that’s kind of used to that digitization anyway. I mean, most people don’t even have textbooks anymore, right? I mean, all the learning is online. And so how do you translate the educational to an adjusters knowledge transfer to the next generation? You can do that very easily, you know, with the way we’re set up, I think, an origami and a Clara to show both systems. I think, you know, what we’re also seeing with that, and Jamie can speak to it even more so, is, you know, I think everyone is talking about there’s the core system, there’s the role of the core system. Okay? But that intelligent layer, we’re getting to that place for so many reasons that we talked about. How do you start to bring those two together and dance together and accelerate capabilities for the claims adjuster or the chief claims officer? And I think Jamie can speak more to that. Absolutely. So, right, origami, kind of thinking about origami as a bit of like the hub and spoke or your core claim system. So we know there’s an incredible amount of data that’s coming into the system and informing whether it be some sort of medical bill review, if there’s some sort of pharmacy benefit management that’s happening, you’re doing your, you know, your Freys and your Shroys. Like, there’s an incredible amount of information that’s being passed in and out during the life cycle of your workers’ comp claim. And so being able to make sure that you can manage all of that data in a centralized place and being served up with the information that you need in the moment that you need it, right? So that’s exactly what Heather’s talking about, especially as we do this transfer of knowledge within the industry, as we start building up the expertise of this up and coming generation of insurance professionals. Like, that’s going to be critical. And not to mention the fact that they’re going to expect it. They have, you know, their lives have been with, in many cases, devices where GPS is already integrated, where it’s by default turned on. So I think that’s the opportunity that we have as more tenured professionals in the industry to help not only supplement what we can do in our work and adopt to some of these new technologies, but to build to the expectations of the up and coming generations. I love how you talked about, Jamie, just the you’re talking about the, like, the amount of documentation around these super complex workers’ compensation. I have so much empathy for an adjuster. Something that we always do as AI professionals is we always shadow adjusters. And so, you know, obviously their core system, but then there’s still things that are, I would say really heavy for them from an operational efficiency perspective. So getting the medical documents in a PDF format and still having to read through all of the documents. I mean, that could be thousands of pages for medical, obviously, stats embedded in there. And I was watching this and my team and I were just like, you know what, is there a way to apply an AI product to this? How do you, again, give them some sort of a second set of eyes? How do you build an AI model that can not just digitize and coalesce and put it in chronological order, all these medical pages on a five year workers’ comp rate and make it searchable just like Google? But how do you surface up things that maybe there’s conditions in there that are so embedded in a hundred page document and their tired eyes at night when they printed out everything? Well, that’s what we do too, right? Both on legal demand packages and medical documents. You know, our AI models are surfacing up to them. Listen, we’re seeing a condition in here. We’re seeing a treatment code that doesn’t really match here or medication or additional procedure or a new bodily part. Like, wow, you know, I kind of wish I had one of those myself around many aspects of life. But these are the type of adjacencies that we really are trying to help the adjuster because so much data doesn’t always mean it’s the right thing. Right? But getting really that surfacing up what matters to them that can move the needle on that case. That’s what we’re trying to do is really help them through that process. Very good. Okay. You know, I I think we’re like, often with with technology, it seems like we’re on a bell curve. We’ve got, you know, early adopters. We’ve got kinda laggards, and we’ve got folks who are somewhere somewhere in the middle. And artificial intelligence itself is kinda continuing to evolve and and get refined. So I think, where is that going to take the industry? And and I’m I’m curious to know how do you see the future of that, and particularly kind of this, you know, this thing a lot of folks are are hearing about is the generative artificial intelligence and maybe maybe other applications of that kind of technology. How do you see that applying in in the workers’ comp field? Well, I think we first need to acknowledge, you know, our audience here and our attendees and where they might self identify and where they fall on the bell curve. And first of all, completely normalize and indicate that it’s perfectly okay if you are kind of early in that bell curve or a middle adopter. Like being here, being interested, asking questions and growing your education around what AI is and all the many things that it means. And you talked about Gen AI specifically, and I will certainly get to that. So just want to call that out as an important, like we’re at an important point in time where it’s really about education right now. It’s really about growing your comfort with it, and it’s about exploring and experimenting so that you set yourself up for the inevitable adoption of the more and more complex models and uses of AI. Specific to generative AI, we at Origami are currently running a beta program with a number of our clients around trying to really understand if our user base, if our adjusters and also like underwriters, if they’re ready to start really working in some of these generative AI, you know, GPS helper into their day to We know we feel it’s inevitable, but starting to really see what are the use cases that they’re willing to adopt? What are the habits that they’re willing to change today? And they’re very simple use cases. They’re things that people can start experimenting with on their own, with like ChatGPT in their personal life, for example, like email summarization or content summarization. Like just what Heather talked about with going through a lot, a large number of documents, how can I start to be able to consume more quickly? How can I create that little piece of efficiency that, remember, these types of things have incredible cumulative impact over multiple people multiple times? The other thing that it’s great for generative AI is fantastic for, especially as that first stage engagement, is gonna be content generation. So one of the things we’re experimenting with is email generation. There’s an incredible amount of communication that’s happening between an adjuster and attorneys and providers and of course the claimant. So beyond just a standard kind of like mail merge or template, how can we create a more personalized experience? You know, I love when Heather talks about, like, empathy for the adjuster, empathy for the claimant. And I think, you know, anything that we can do to kinda even if it is technology assisted to create a more personal connection is gonna be really helpful and is a great way to start exploring. Yeah. Just to piggyback off that, I would not answer that, Regis said, in three ways. So on the first one, we do have all across the spectrum that early adopters that started with us back in twenty seventeen, twenty eighteen to, you know, ones that now really have their AI claims playbook strategy and want to jump into this with an AI claims platform. So that’s beautiful because we actually have wonderful case studies of ROI from from the early adopters that can share really good experiences with an AI platform. And we only get better and better. I think on the GenAI front, you know, we love to build these derivative products around summarization. So we know the adjuster, you know, if we’re prompting them to go into a case, the models are detecting something around the injured worker in the case. It could be a lot of things, you know, propensity to litigate. It could be return to work is being elongated, whatever it is. And they go in, you know, to have a have a claim summarization and have some cliff notes there and bring them back into this movie of this case and with the associated documents of like the last ninety days and where it is in the claim cycle, that’s a nice usage of Gen AI modeling and to be able to give them a baseline back into a case instead of really having to go deeper and deeper and deeper. The second one that we really see gain a lot of traction is around, the Gin AI summarization of all of these documents. You know, it’s just an efficiency play for an adjuster. Right? One, you know, I think a lot of people just think digitization is just sending a PDF, you know, manually. But, we’re able to incorporate it as a part of our other products too. So the medical documents coming in, we’re sensing something, then how does that then it’s not just a standalone. It can be. It’s an efficiency play for independent medical evaluations that can be a side eyes to these physicians that are working in that field for someone. But it also becomes a part of the bigger treatment case and what’s going on here. And then is that helping the case from some type of triaging of it? You know, we are a platform that all the data really works with the other products to make them and strengthen them, you know, even more so. So, you know, truly believe there’s so many additional opportunities when you sit down with an adjuster of of helping them with efficiency and additional AR products that that we’re gonna bring to bear here. Good. That’s actually a really good segue to, you know, another another question, which is, you know, kind of from a tactical perspective. You indicated that your data doesn’t have to be perfect if you wanna implement AI and get some of the efficiencies in in the claims function for that. But what are some of the kinda next steps that insurers should be thinking about taking when they wanna harness AI? What are what are kind of the things they need to do to get ready for AI even if the dataset doesn’t have to be perfect? And and how do you see them kind of thinking through their application set to, you know, to make the best use of of AI and and what you can offer? I would say become a champion for AI within your organization. So we’re hearing from some organizations where that push for technology and specifically some operational efficiencies or even kind of product efficiencies are coming from a top down leadership. But as I think people are hearing here in this conversation, my belief is that this is inevitable. This is coming. This is going to offer incredible benefit both for individuals and for entire organizations. And so doing that education, leaning in, experimenting in personal life if you don’t have full support from your organization yet, that’s where it really starts. Yeah. And I would say, just like Jamie was saying, it’s here and it’s going to stay because it really is going to be that GPS of of claims and how to guide these cases. I think it’s gonna be a competitive edge, which translates, you know, back to the policy side and capability as they look at pricing and so forth, one is going to help the other side because a lot of times we get asked, okay, you have all the claims data, how can we give a heads up back to the underwriting side, right, And help with those triangles that the actuaries or pricing are looking at. And it’s very easy for us to obviously, we’re an API architecture, and that’s why I work so beautiful with Origami with integrations, is to send that over to the underwriting side and really close that feedback loop and help with those signals from pricing to with claims. So it’s a constant with with with that piece of it. Okay. Good. I think, you know, the fact that we have you here together both representing, you know, two leading providers in the technology space, it it seems like it’s it’s a a natural combination to have core technology, you know, a core a core system with artificial intelligence, you know, and and and the analytics piece. So how how do Claranalytics and Origami Risk kind of work together? How what what does the collaboration look like for both of you in terms of integrating this new and exciting application of of artificial intelligence? Go ahead, Amy. Yeah. Well, first of all, it starts with, I think, you know, the conversations that Heather and I have as well as our teams have from, like, an organizational partnership perspective. Like, we’re very clear on the ways that we help each other and then are in turn, most importantly, able to help our clients and in particular our shared clients. The other thing that Heather talked about already is that integration. So in the ability to integrate. So Origami natively does is able to very easily integrate with many different systems and be able to pull in data and export data out of the system as needed. And so that’s the critical piece there is that partnership and the technology that’s in place that allows us to easily integrate and do so with speed. We always love when when when we’re speaking to someone and they tell us we’re on origami because we’re like, good. And what that means for us is not only the respect we have to what Bob and colleagues have built, but So architecture, so many of our clients will say, our data is in seven different places. And we say, great, just tell us those seven different places and we will hook up the pipes literally, right? To start the extract, the ELT piece of it. You know, we don’t have to do all that when we’re dealing with Origami, right? Because we already have that one pipe that’s coming through with them and that beautiful integration. And so we do see though a lot of our origami clients now already getting that playbook for AI and their nomenclature and really thinking about at the board level in their claims organization. I got to go there next. Where am I going to start? So I think that’s one thing. We’re glad that they’re thinking that way. I think back to the piece about the data, as long as you have data, we’re good. Okay. We can work with the quality of it because sometimes it all starts with entry. And if we see that there’s some fill rate problems or some integrity issues, it’s a full feedback and we’re able to push that back and help them back into, let’s just say an origami system. So it’s a nice, beautiful, we give them a data quality index from like, this is the threshold where you are. We only need eighty attributes for this particular model. Eighty, that’s it. We don’t need all that. And so we’re able to work with that. But we transform it, we map it. We just need some data. We just don’t care where it is. We don’t care the form that it’s in. That’s our data engineering recipe of what we’re being able to bring. That’s great. Well, I I think, you know, the the question here really is how do you get the most out of the technology stack that you have? And and how how are Origami and Clara clients seeing that? You know, what are what is there is there a formula to kind of get the most out of the systems that you’re using? To me, it’s the commitment to continuous improvement. It’s the commitment, and that commitment needs to come from the leadership within the claims organization. The expectations of their leaders and of the adjusters is really the commitment to technology, the commitment to the way that technology can improve these operational efficiencies we’ve talked about in this call, and then ensuring that you’re looking to always get a little bit better. And remember, there’s big ways to do it. I think exactly what we’re talking about with Clara. Those are really big impacts that we see right away, and then there’s small ways to do it. There’s just little tweaks to the way that you are navigating through your core system or the way that your core system has been configured. So not just settling in on how you initially implemented something and knowing that it needs to continue to evolve. And that type of mindset sets you up well for this transition into this AI future. Yeah. Each is needed and each has its respective role. And we don’t do what the core system does, and the core system doesn’t do what what we do. And but with the way that things are are moving, you know, more and more people, I think, I wanna speak for Jamie, at the core system are asking where where is that AI intelligent layer? When can I have that? What does that look like? Can I start to tippy toe into it? You know, because I I have a little bit of trepidation. And that’s the beauty of us being able to work with Origami is that we can work with those clients how to, you know, gingerly, gently get into their AI playbook and quickly see with our UI and the core system UI, how these two can be married up and each has its own its own role. And that’s why I always talk about the three screens because it’s like you got the Zoom, I’m going talk to the claimant. I got my core system. It’s doing all that administrative and some of those insights. And I’ve got this signal over here, my GPS, my third screen. Mean, it’s a full cockpit here. Right? Think about the adjuster. I got a full cockpit. And then the fourth thing that we talk about is air traffic control within the tech stack is listen, so many super complex cases need a little bit just like with airplanes, right? I’m at this altitude and I need a little bit of help. There’s prompting that systems can do to help a supervisor or chief claims officer understand through any type of communication protocol that you want through our app to be notified when something’s happening on a certain set of cases that have a really high reserve and high severity. Come in, come in and help the pilot a little bit here that’s driving these cases. And so that visibility right now, if you’re doing all this manually, it’s kind of like, know, listen, I’m behind the eight ball here on this case. I’m just seeing this versus I’m being preventative and proactive because the model is telling me this about the case and I don’t want this outcome to happen. And these are the three ways because generative AI, you know, because machine models have learned from Clara for the last ten years. So the models are like, okay, Heather, you’re the adjuster. These are three different ways that this case could end with where it is right now. From a cost, from an outcome, okay, or from a settlement, from a litigation case because our models have seen all the different tree branches of this, if you will. And so to have prevention and be able to steer the case not from, I’d say, the way the traditional model, but the new paradigm of having that heads up all the time. I think that’s changing to the way that you’re able to lead your claims organization. And, you know, just like, you know, when you fly planes, they’re like, listen. A hundred miles up, we’re gonna have some turbulence, and I’m gonna be pushed I’m gonna be asking you to put the seat belt on. Okay? Think about an adjuster that’s like, listen. I’m getting some prompting here and able to tell air traffic control, clean supervisor, we might have some issues in this case. And this is the different ways. It just changes the conversation to you. Right? Not on your heels. Yeah. I think what the the theme there is finding the trusted technology partners to help you on your journey is critical. Yeah. See, that’s why I need Jamie. She just, like, brings it all. She just lands the plane on me. Lands the plane. That’s good. You know, a little earlier, we had referenced one of one of the case studies, you know, as demonstrating return on investment, you know, in in this technology. Is there is there another one that you wanna talk about in terms of that really kind of brings to light the ROI for for insurers? Yeah. I think we have one, in here. We just if you wanna if you wanna go there. Yeah. So this is a, this is a workers’ comp, carrier that started with us. It was a really nice onboarding. So as I said before, five years of history of of claims data we received so we can baseline to understand how the claims organization has treated those cases so that we can give attribution to the models and incrementality. And so after six months of implementation, we started to look at the ROI and agreed upon with the carrier of the Clara models contribution to these costs. So you can see here, you know, eighteen ks volume self administered. They really wanted to use, as I talked about that provider products to optimize their network. How can they really just only have a, b, and c scored? Some of our clients take out c scored and have objective data around the providers that have been treating their workers’ comp cases. In our application, you get two scores. So this particular care would get the score of the physician and their respective data set, their ecosystem, but then be able to benchmark it against the industry score. So we have a contributory database, which is anonymized. We don’t take in PII data so that all of our clients can benefit from new physicians coming into the network that maybe is scoring at a certain A, B or C that can then add to their scoring repository as well. The same on the defense panels and litigation panels too. Anyway, twenty nineteen signed it, started really looking after six months at the ROI. We have former actuaries on our staff that do this work. And you can see for yourself on the right hand side, using that treatment model, really getting the right physicians in their network, getting rid of their D and E score physicians, and then also triaging these cases in a different way for them to make decisions has allowed us to and a couple of these stats actually were in their annual report allows us to give back to them this nice ROI based on truly, truly a nice ACV to this ROI that, this chief claims officer, loves to talk to, the CFO and CEO. But a really high adoption, on the alerts. We sit down with them all the time. We allow the adjusters to this is their Clara. Right? It’s not my Clara. It’s your Clara. I want this to work for you. And they help us with the road map. They tell us additional functionality and other alerts. But this is just one example of many where the AI is producing some really nice financial returns. Obviously, there’s operational returns too. If there’s a way that we can lessen the assignment of the segmentation, the turn. So one of our clients, it was the assignments to the cases were turning almost ninety eight percent. We brought that down to two percent. So really getting that assignment of the case to the adjuster within the first or second, hopefully the first is what we’re doing now for our clients. I think really the accuracy of the reserving, really getting that down, not taking twenty one days, but seven days, just another, you know, a lot of just different operational metrics too beyond the financials that we really see. And obviously, you know, bringing in the closing days of the case, right? The duration of the case, really helping with those things. There’s a number of metrics that are compiled underneath this. Okay, good. Jamie, anything to add to Hathrich’s remarks? Not on that one. Thank you. Well, I think we have a few minutes left in our time together, and I’d like to open that up to the Q and A portion. The audience has submitted several interesting questions here, and one of those is where do you see AI kind of having the most impact or getting the most utilization perhaps in workers’ comp claims? You know, some things we talked about earlier is, you know, certain certain outcomes Yeah. Prior management, you know, the operational efficiency. Do do you also see it taking place in in a bill review, utilization review Yeah. Things on, you know, management level feedback on the claims? All of those. All of the above if that was a multiple choice test. And you passed. One is because our models look for same like claims. So I’ve seen that spinal injury. I’ve seen that orthopedic injury of something that’s happened in the warehouse two hundred times. So our models can quickly, because they’re looking for the same light claims, say this is twenty weeks. This is a twenty week from the time first notice of loss choosing an A, B, or C provider out of Clara’s treatment product, getting the right one, here are the procedures, the codes that we would see in this type of injury, and after physical therapy, and then some rest in about twenty weeks. We chart that. Just like you chart driving from A to B, we chart that. That is the goal. And the outcome, okay, so you do not incur lost costs, and we get that injured worker back. Stuff happens in the cases, Right? And that’s where you’d start to get the alerts on some of this. It’s not gonna be twenty weeks. If you don’t intervene, it’s gonna be twenty five weeks, all that kind of stuff. Right? I think that’s one case. I think the second case is I think Gen AI is gonna continue to provide summarization where we can. Okay? Think about a cliff notes version of that case. Where can we help you the most to get up to speed and back to the movie of that case? Because you’ve been handling one hundred and fifty other cases. Right? And then how do we help you on all those medical and legal documents? I mean, these legal demand packages with subconditions underneath of it. How do we alert you to some of these things? We’re not doing your job. All we’re trying to say is speed bump ahead, traffic ahead, accident ahead. We’ve just detected these things. So I think we’ll continue to see more opportunities to, I think it’s only gonna get better and better. Our models keep on getting more sophisticated through understanding natural language. Listen, can seem when there’s a duplication of something, of a document. Okay? So I think that you know, we’re just going to keep on training them and look for ways in partnership with claims organizations to help them. The place that I’m watching right now is actually on the personal line side. So observing what’s happening, like, the personal auto insurance industry and the way that, like, personal insurers are interacting with their customers and, like, within the claim process. And I think one of the places that we’re gonna see pretty fast impact is on intake. So ease of intake, making sure that we’re getting information in a really effective and properly categorized way. And that might happen through AI chatbots. That might happen through even some sort of imaging or videos that can happen. So I think that’s one place where there’s gonna be some fast impact that as a result is going to have the ability to very quickly allow adjusters to make plans and fast interventions. And I think to piggyback off that, I think there’s some great learnings on the personal lines from an adjuster to bring over to commercial lines. So when I think about other industries where let’s just say after first notice loss, there’s step one, two, and three. How can you use AI where the human doesn’t have to take some of those questions, where the human doesn’t have to interact? But when it starts to get really complex, when you’ve got to really talk about the provider or talk about the case and other questions, how do you architect some of that to help the adjuster and that AI platform, these initial things that are coming in even on a commercial case? I think there’s lessons learned from personal to commercial. Good. Several of our audience members have asked another interesting question, which I don’t think we really got the chance to touch on, and that’s utilizing this technology to flag fraud, you know, where there’s a potential, you know, fraudulent claim in workers’ comp. And as we know, unfortunately, that that does happen. It’s it’s not certainly a majority, but, you know, given the universe of workers’ comp claims, you know, even a one to two percent, if there’s fraud in that amount, you know, multiplied, that’s that’s a big number. How how is this technology looking at kind of flagging fraudulent claims? Yeah. So we’re asked that probably every day by our current clients. You already have all of our claims data. You know our relationships with our provider network, and you also can see any type of relationship of that provider with someone in the legal field as well. So it would be very easy for us on our platform to turn on a capability of looking at fraud. We’ve been asked to do that. It would be a nice adjacency capability for us to turn on. So it is something part of our roadmap that we are considering because, you know, we really wanna be that kind of headless app and working with the core system, but integrate whatever type of, you know, hybrid model, alternative datasets even outside of our datasets, right, that we’re getting in and in other type of capabilities, you know, whether it’s a model built inside the carrier or self insured or externally. And so we do think fraud is something that would be a nice adjacency because we’re being asked to do that. Very good. I think I think we really have time for maybe only one more question, and, and that’s that’s another one that’s been on the minds of our audience. And that’s, what kind of guardrails should the industry be thinking about in terms of utilizing the data, you know, the sensitivity of of the data? There’s certainly opportunity for anonymization. But how how do origami and Clara think about, you know, the the guardrails around safeguarding the the the data of claimants? Yeah. I think like you said, we’re not we’re not using PII data. We just need the injury, right, and the location and the industry. So we really also make sure that we are surfacing up options based on what the alerts are seeing because the adjuster, the human in the loop, at the end of the day has to make the decision. We’re surfacing up things that we’re seeing and here are potential outcomes. Know, here’s three tree branches based on what we’re seeing what we’ve seen with other cases. But we make sure that that’s why we need a human in the loop. They’ve got to make the ultimate decision. We’re saying here’s your options. Here’s what we’ve surmised by the last ten years of learning this, and we’re not using PII. So, you know, that’s where we really keep the guardrails reach us. Yeah. And there comes the human in the loop is totally critical here because you need to have that education and the awareness of where the potential current pitfalls, particularly on the Gen AI side are, like with the hallucinations, the potential biases in the models. You know? And we know, you know, I’m sure Heather and I and many others are watching the regulatory, like, impacts from around the world around expectations on AI modeling and how that’s gonna impact all of us in the way that, you know, we may have to uncover sources and the algorithms and whatnot. So having awareness into all of that, and in particular, being very clear about the data that you’re sending in. If it’s a generative AI, you need to be very conscious of putting any sort of protected information into your prompt and understanding how that model is utilizing that information. So that’s been an active conversation within Origami. We wanna make sure that our clients have an understanding of what they’re choosing to interact with and that there’s that shared education that we’re doing our part to protect and to educate so that you feel confident in how you are interacting with these technologies. Terrific. Well, unfortunately, we’ve come to the end. I think we could continue talking about this for the rest of the day, certainly. But I’d like to thank you, Heather and Jamie. Thank you for being our panelists and giving us a great conversation on this topic. I’d like to thank our audience also for tuning into this and for your good questions. A replay of this will be available probably in about twenty four hours to all who registered. You’re welcome to, to invite a colleague who wasn’t able to make that, to see the on demand replay over the webinar. And, for more information, please download a white paper on artificial intelligence at origami risk dot com slash workers’ comp a I. Really appreciate, your opportunity to, to learn with us in this webinar, and, we hope you’ll, you’ll join us for future ones. Thanks again, everyone. Have a great day. Thank you.