[00:00] Welcome to LenderDock Unlocked, the podcast that delves into the revolutionary technology behind LenderDock, the leading provider of online property and casualty insurance policy verification and automated lienholder process management services.

[00:14] Our policy verification as a service platform offers banks, lenders, and financial third parties the ability to digitally verify and correct home and auto policy related data in real time.

[00:26] Join us as we explore the ins and outs of this game-changing technology and hear from industry experts on how LenderDock is streamlining and simplifying the insurance verification process for all.

[00:37] Tune in to unlock the future of lienholder automation with LenderDock.

[00:48] Welcome to another episode of LenderDock Unlocked.

[00:52] I’m Frank Eubank, CEO, and I am joined again by a, a very good friend and a great close partner, Ben Bomhoff.

[01:01] Ben is, as was previously introduced, is the Chief Technology Officer for Fleming Insurance Holdings, and, we are really excited to have him back, and we’re gonna be talking about, a really great topic that is, we, we feel is very on point with, so much of what’s being talked about in the industry, and that’s- That’s, AI and how that’s being used and leveraged within the industry.

[01:25] So Ben, welcome to, to the call today.

[01:27] Well, great, great, great to be back with you, Frank.

[01:30] I, I appreciate it.

[01:33] I, I think, I’ll just tell you a little bit about, about myself.

[01:36] I’m the Chief Technology Officer for Fleming Insurance Holdings.

[01:40] we’ve, we’ve had a great relationship, Frank, in, in working with LenderDock and with other companies, and so I’ve, I’ve always admired, Helped out the businesses I’ve been associated with, and, I’ve been in the business for about forty years, over forty years, in, in property and casualty insurance the whole time, so, been around the block a few times and, that’s, That’s kind of, my background.

[02:10] But, as we say, you, you, you’re a longtime, tenured expert in the industry, and, and, you know, it, it, it’s– that’s something we’ve always appreciated, is just the, the amount of experience and insights.

[02:20] And, and I think that kind of leads us into, you know, really what we’re wanting to hopefully cover today, and again, this, this broad topic, of, of AI and how it’s being leveraged or really how the industry is trying So I, I, I wanted to maybe turn some time over to you to kind of just give some insights or, you know, provide our, our listeners, you know, what you guys are doing because there’s some really cool things.

[02:47] So.

[02:47] Yeah, yeah.

[02:48] So with, with Fleming Insurance Holdings, I’ll tell you a little bit about what we do.

[02:52] We’re a runoff reinsurer and an ILS platform that’s Insurance Link Securities platform that provides capital optimization and liquidity solutions to the insurance sector.

[03:02] So that’s a bit of a mouthful, but what that really, in, in regular Let’s do an example where we have an insurance company or a captive that’s written, a, a comp, a work, a work comp book, and of course, the tail is long, ten years, however long it goes, and there, there’s a lot of capital tied up in that, in those later years, that, in that tail of that insurance.

[03:25] So, there’s, there’s IBNR, there’s reserves, there’s operating expenses all tied up in running off those last, say, seven years of, of the ten-year tail that you’re looking at, and The offer is, is to take that off the hands of that re- of that insurer.

[03:42] That frees up capital, they can take those reserves and put them into more strategic, functions in, in their company and, Give them, also give them some finality in their, in those, those losses that we take over.

[03:56] So we literally put it on our balance sheet, we take responsibility for those claims, there are liability.

[04:01] So that’s, that’s what we do.

[04:04] We need to make sure in our business that, we, we have, keep a close eye on those claims, we, we make the outcomes as less severe as possible, that sort of thing.

[04:14] So data is a big component of our, of our business, and, we, we see You know, AI and ML, AI and machine learning.

[04:24] And by the way, Frank, I, like I said, I’ve been in this for forty years, I can’t ex- I can’t remember a more exciting time to be in technology in any industry right now.

[04:34] This is just an amazing time, fascinating things going on, so, and, and so much opportunity.

[04:40] Yeah.

[04:41] So yeah, no, I, I, I couldn’t agree with you more.

[04:44] And to your point, right?

[04:45] it, it’s, it’s so, things are picking up and they’re accelerating at, And we all know that, you know, within the insurance industry or that space, the, the, you know, our industry’s been a little bit of a laggard when it comes to, you know, forced adoption and trying to digitalize or modernize kind of the infrastructure or the architecture of running the business.

[05:07] And so, you know, we see a lot of that going on with just new, you know, policy admin platforms and new operational services.

[05:15] But, you know, w-with what you’re seeing on your side of the world, maybe tell us a little bit about what you- You guys are doing with AI and kind of where you see that sort of supplanting itself as a value?

[05:27] Yes, yes, and, and particularly, really fascinated with the, the neural networks and the other cognitive learning, as well as the, the regular generative AI components, really fascinating things, and, and we see some, some use cases that I’ll, I’ll, I’ll talk to you a little bit about.

[05:44] also the, the, the amount of unstructured data that we have in the insurance business is also huge and is- Is a key component in being able to leverage AI and ML and using AI and ML to get at that unstructured data and get, you know, good data points and, and trends and things from those, from those unstructured documents.

[06:06] Right.

[06:06] So it’s, you know, that’s a, that’s a big challenge, I think, in itself, just, mining, mining the important things out of those, those, file attachments and, and documents.

[06:16] So some of the use cases, you know, I described our business and some of the use cases We’re seeing that are gonna be very relevant to us is due diligence on new deals.

[06:25] When, whenever we do a deal like I described, where we’re gonna take a, a book of business off of someone’s hands, we need to really have a look at it and see, okay, well, what is it?

[06:34] How do we price it?

[06:36] is it properly reserved?

[06:38] you know, ’cause we gotta make money on these deals, so we, we wanna make sure that the, we price it right and fairly to both parties.

[06:44] So using AI, on– we So you have a, a large data exchange, and what we wanna be able to do is leverage some of those tools to look at the history of that book we’re doing, not just the open claims, but the closed as well, and be able to take a look and see, you know, where, what, what are the patterns, how do they reserve, do we think it’s reserved properly, those sorts of things.

[07:09] So that, that can be a critical tool for us to get the right deal or maybe even walk away from a deal if we had to.

[07:17] So, that’s After we, after we consumate a deal, we want– obviously want to reduce the severity of the claims.

[07:25] They’re now our claims that comes off of our, right out of our bottom line, right?

[07:29] So we want to be able to, use AI and, and, and machine learning and predictive modeling to reduce claim severity.

[07:37] There’s plenty of opportunity in, in claims that are in litigation, which we ha– we’ll always have a lot of, looking at things like jurisdictions, who’s the opposing attorney, who is the judge in, in historical cases can help Help us understand what and, and who was our attorney, did we lose, did we win, and for cur- for applying that to current claims, so you know we can

[08:00] A claim that’s in litigation.

[08:01] the, the da-the, the data tells us that this guy always beats us, this judge always hands off, you know, Decisions that aren’t very favorable to, to the insurance business.

[08:15] And so the answer might be to the claim, you know, to the people handling the claims, hey, let’s just get this thing settled, let’s not spend another million dollars in, in legal expenses and then lose it then, right?

[08:26] So, wow.

[08:27] Yeah, that, I mean, gosh, that makes a lot of sense.

[08:29] You, you talk about, so we talk about big data, right?

[08:31] I mean, and that, that’s, that’s a coined term that we all are familiar with, but maybe tell us That, right?

[08:41] Handwritten notes and, and claim documentation and, and even, even imagery and photos now.

[08:46] So w-w-where, where does that all fit?

[08:49] Yeah.

[08:49] That, the– and that type of data, I think that’s where the real gold is in, in, in finding out what’s going on in, in a claim file or, or a medical file or any, any type of, dataset, i-i-are those things also in, you- So they usually take the form of file attachments on claims.

[09:08] So you’ve got medical reports, you can have police reports, you can have surgical, reports, all kinds of things.

[09:16] You also have adjuster notes that are basically just free-form, free-form notes as they go along.

[09:22] Hey, here’s what happened.

[09:23] Hey, I think we’re going into litigation, I recommended this.

[09:26] a lot of the things that you would love for claim systems to have as discrete, structured fields, even if they’re there, people don’t use them.

[09:35] Most of the time, so all that, all those important data points are buried in unstructured data.

[09:40] So that’s a key, a key thing for us, Frank, is to be able to break into that stuff and be able to see, see what’s going on.

[09:49] So, there’s, there’s some tools, some, some, artificial intelligence that, that really addresses Very specifically that, that component which we’re really interested in.

[10:00] Just a couple more use cases.

[10:02] Yeah.

[10:03] we run into a lot of complex claims, the older they get, usually the more complex they get.

[10:07] They can have four hundred, four hundred file attachments.

[10:11] how do we get a handle on what’s going on with that?

[10:13] We’re seeing some technology that will sumar– do summarizations of a bunch of different disparate documents and, and put together, okay, here’s what’s the state of this claim, what’s the health of this person who’s, who’s been injured, that sort of thing.

[10:29] and also, we can also validate claim accuracy in, as far as the claim handler goes, by comparing structured data like cause of loss, those types of fields, and does that corroborate with the handwritten notes?

[10:44] there’s, there’s that level too.

[10:46] So talk about, you know, quality of, of, of claim handling, we can do that on due diligence as well as on, as we’re, we’re, we, assume that business as well.

[10:56] But those are some use cases we’re really interested in, and I think we have some great opportunity, to do that.

[11:02] So how, how, how accelerated is, is that for you and Fleming?

[11:05] I mean, what, what, what kind of, where in the stage of, of real production application do you feel that, Yeah, kind of, you know, our approach, we’re, we’re, we really started seriously, with this probably about, six to nine months ago, where we, we are literally at Fleming, we’re still getting our, our data, our data lake and data warehouse together right now.

[11:31] So we’re putting all the infrastructure in that can handle, machine learning workloads and, and AI and, and, predictive modeling to do that.

[11:41] So what we’re doing right now is we’re- We’re having a lot of meetings and a lot of demos with, with pro– dozen, dozens of some of, of these companies.

[11:50] There’s so many out there.

[11:52] our approach is we’re, we’re in the, that phase where we’re evaluating, and we’ve got a, got things narrowed down pretty, pretty, pretty good right now.

[12:02] one of the things we found too is, is some providers give you kind of a soup to nuts where, oh yeah, we’re, we’re great at unstructured data, we can pull that for you, we can then move Models, we can do predictive modeling.

[12:15] We’ve got a big corpus of old, you know, anonymized claims, so we can, we can do historical, predictive modeling, that sort of thing.

[12:23] And so, you know, there’s that.

[12:25] Then we’re also finding some smaller firms, a few in Silicon Valley, some startups we’ve been talking to that specialize only in getting an unstructured data and getting out of that what you need.

[12:39] And so what we’re finding is there are some companies There are different companies that do things really well, and, and, and so that’s what we’re evaluating right now.

[12:50] I, I think there’s, we’re probably gonna end up with a combination of tools.

[12:55] And for example, there’s one company that can pull out information, do summarizations of, you know, from four hundred documents, give you a summarization, but they say, “Well, we’re not in the modeling business.

[13:08] Our job is to, to be the best at pulling this stuff out for you and getting those data points and those insights from the- The data.

[13:14] We don’t do modeling.

[13:15] Now there’s companies that do modeling too, so I think we may end up with, a combination of components that are kind of best of breed for what they do.

[13:26] And so that’s been an interesting insight for me.

[13:28] I’m learning so much through this stuff, Frank.

[13:30] It’s really- For sure.

[13:31] Yeah.

[13:31] So much fun and, and really interesting.

[13:34] I’ll just give you a quick- a-and not only that, the, the, the technology is emerging so fast, you, you’ve- I get worried about getting locked into one One technology, you know, because somebody’s gonna come up with a better mousetrap real soon, I, I think.

[13:51] Yeah.

[13:52] Tomorrow.

[13:53] Yeah.

[13:53] I will, I’ll, I’ll digress for a second here, Frank, i-i that back in, late, the late, like around two thousand and seventeen, two thousand and eighteen, with the company I was with, we created an, a chatbot for the insurance business.

[14:08] And what we did was there were some great tools, IBM Watson had some great stuff, AWO- US had some great stuff.

[14:15] You know, one, one product was really good at speech to text, the other one was a little bit better at voice recognition, the other one had better, you know, inference, inference analysis and sentiment analysis.

[14:27] So we said, you know, How are we gonna do this?

[14:29] So the first thing we built was an orchestration layer where you call the services to do certain things, so that way we weren’t hardwired into the speech to text.

[14:40] You know, it could change.

[14:41] The, the new one’s gonna come out.

[14:43] And, and now, now we’ve, we’ve got kind of a platform that is loosely coupled so that we can do that.

[14:49] So that’s kind of my mentality with AI right now.

[14:51] And whether that plays out right or not, it’s all so new, we’ll see.

[14:55] But that’s kind of my thought process.

[14:57] Okay.

[14:58] Tell me, so I was gonna ask you, just another question as I’m thinking about this.

[15:02] What are the risks, right?

[15:04] A-and, and meaning that might be one where There, there’s three other technologies that are just better or further advanced.

[15:13] But I mean, what, what are, what are in your mind, what are the risks with AI right now?

[15:16] Well, and, and that is, that is one of the risks if, you know, if you’ve got, if you’re into one pack, one package and one platform, you’re dependent on that one.

[15:24] So if somebody really blows one function out of the water like, like really analyzing unstructured data a lot better than most, then you’re kind of stuck in that.

[15:35] So I think it’s a Infrastructure to make sure that you can plug in the best of breed, it may be one, one product, it could be, but if it’s not, it’ll allow you to do that.

[15:49] I think the other thing is, is making sure we get proof of concepts in place.

[15:53] Again, this is- Especially with claim information, how do you prove that you handled it better?

[16:00] Unless you handle it the old way and the new way, right?

[16:03] It, it’s, it’s very, it’s, it’s hard to– it’s, it’s not like using AI to reduce someone’s workload or take over a call center or things like that and free up people’s time, those are easy to quantify.

[16:15] That with this, you know, a claim outcome, well, you don’t know what the claim outcome would have been You know, had you used this or had you not, because you can only use, you can only try that claim one time.

[16:26] So that’s right, yeah, that’s a point.

[16:28] So that’s a trick.

[16:29] interesting.

[16:30] So, so the, those are the risks, Frank.

[16:32] I think you can invest a lot of money, but you gotta make sure your, your use cases and your business case and your cost benefit is gonna work, because it, it can get expensive really fast.

[16:44] Yeah.

[16:44] And, so I, I think it’s there, you can’t be afraid of it, it’s there, and it’s coming.

[16:50] It’s just a matter of, I think, how, how we all decide to implement it.

[16:56] Yeah, there’s, there’s, there’s, I don’t wanna see a lot of noise, but there’s just a lot of attention, a-and, and it’s coming from a lot of different directions.

[17:02] And so I think, you know, listeners, insurers, carriers, providers are, are all kind of looking at this, you know, ears Yeah.

[17:17] And I, and I think Frank, you can find anybody to tell you what you should do.

[17:20] I don’t think there’s any, any replacement for getting out there and figuring it out and getting in the middle of it yourself.

[17:27] I think that rolling up your sleeves and– Right.

[17:28] That’s right.

[17:30] Get your hands dirty, you know?

[17:31] And, you know me, I’ve always gotten my hands dirty no matter when or what.

[17:35] Oh, no, no doubt about it.

[17:36] Some other use– You have some other use cases that, that you, you thought about as far as, yeah, what authorizations, it’s, that’s really important because if we’ve got some claims that have some big potential losses, our claims staff is gonna go and take a look.

[17:57] Now, I’ll give you a real example.

[17:59] We had one claim, very complex, very ongoing, and there were over four hundred file attachments.

[18:06] And, geez.

[18:07] And, and so, and, and- Reams of adjuster notes and, you know, an occasional report here and there, but nothing, well, what’s the state of this claim right now?

[18:18] Where do I start?

[18:20] so some of the tools I’m seeing are specializing in summarization of unstructured data, and they, they’re training it to do, you know, claim specific, also medical case specific.

[18:33] So for workers comp, what’s the, what’s the medical condition of this person right now?

[18:37] What’s his prognosis?

[18:39] What, you know, all those sorts of things.

[18:40] So, that would get our people up to speed on the critical claims faster, ’cause it could take, you know, when you say file attachments, just think of the physical- Thing you have to do, you gotta pop five, four hundred file attachments and then read them and absorb them and figure out what it means.

[18:58] By the time you get to the hundredth one, you forgot what the first one was for sure, and, you know, it could get- Yeah.

[19:04] So, so I think there’s a huge opportunity, not just for us, but I would say for any insurance company to get that summarization, logic in place because claim handlers switch, right?

[19:16] You’re gonna, claims get transferred to other claim handlers.

[19:19] How do you get up to speed quick on that This stuff.

[19:21] And so I think those are some, that, that’s a big opportunity, outside of what we’re doing specifically.

[19:29] Yeah, that’s right.

[19:30] Well, you know, one of the things that, we, we bump into a lot are just all of, you know, the guide wires of the world, you know, that, that actually these, these enterprise level, you know, carrier provider sort of solutions that, that there’s no doubt that they’re thinking about it, but it, it just, the, the question remains is, again, how does it Deploy it to where it actually is meaningful and, and, and makes a dent, makes, makes, makes it, makes it work.

[19:56] So yeah, yep.

[19:58] And I, and I think, you know, we’re, the good thing about rolling up your sleeves and getting out there too is you start building relationships with people.

[20:05] Right.

[20:06] And that’s really important.

[20:07] So we’ve, we’ve built some really good relationships and, you know, we’re getting agreements to do, to do proof of concepts, you know, for not much cost and sometimes no cost because we’re And that’s important, we get a chance to, to see what this stuff does, and, and with working with multiple vendors, we can even, you know, put the same inputs to two different, two different products and see which one does better, we can do a lot, a lot of different things.

[20:35] So I think that’s, very important when you figure out your approach.

[20:41] It, it’s so easy to get con- to get overwhelmed with things in this- There’s, there’s, there’s no doubt about it.

[20:47] And I think what you provided today was just sound bites, and that is, you know, you gotta just start, and you do have to get your hands dirty, but, but that, that, that evaluation process, it, you know, what, what comes out, if nothing else, are relationships, right?

[21:01] That, that you leverage and, and, and will only, you know, sort of better define what direction you go, and, and it kinda helps to maybe simplify that, decision that you end up arriving to at some stage.

[21:13] But, you know, as always been, I, we certainly appreciate Our end, but, you know, to, to, to, to our listeners out there, I, th-this is, this is really exciting stuff, a-a-as we’ve discussed already.

[21:29] It’s moving very fast.

[21:30] And so, you know, I think from, from our perspective, as is Ben’s, it’s, you know, paying attention to it, i-i-is something you almost can’t avoid.

[21:39] So, so again, really appreciate the insights, and, and it’s always been, you’re, you’re, we’re just- Always a welcome guest here at LenderDock, so I, I do appreciate that, Frank, and thanks for having me, and, look forward to, other, other things in the future.

[21:54] You got it.

[21:54] Thanks so much for your time, and, so this, this is it for us for LenderDock Unlocked.

[21:59] We’ll, we’ll, we’ll catch up with everybody on the next episode.

[22:01] Thanks, and have a great day.

[22:14] Thanks for listening to LenderDock Unlocked, the future of lienholder automation podcast.

[22:19] To learn more about LenderDock’s suite of products and how we can help your business, visit lenderdock.com, that’s l e n d e r d o c k dot com.