When AI Meets Truck Insurance

Raymond Moss: This is the hardest part, 'cause I'm waiting for a funny joke.

Ben Curtis: Why are you waiting for a joke? Because of how we started last week's episode?

Raymond Moss: Podcasts start with funny jokes.

Ben Curtis: Well, you I did have a conversation recently with somebody. It was humorous. Want me to tell you that?

Raymond Moss: Who I do. What was it?

Ben Curtis: Well, so story goes that there was this insurance company, they had hired an AI firm to build a virtual customer service representative.

Raymond Moss: I bet it went really well.

Ben Curtis: Yeah, after months of development, the executives gathered for the final demonstration. The project lead explained to them, all right, you'll be playing the role of a motor carrier. Since your last renewal, you've had two crashes, including a fatality. You've picked up a new basic alert for unsafe driving and vehicle maintenance. So your policy just renewed with a 42 % premium increase.

Raymond Moss: Boy.

Ben Curtis: He smiled confidently, go ahead, ask it anything. So one executive steps up to the keyboard, types, why did my premium go up? The AI replied, that's an excellent question. A premium is the amount you pay for an insurance policy. and can be paid with various forms of payment, either monthly, quarterly, semi-annually, a few paragraphs later. Based on my analysis, your premium could have gone up for several reasons, including but not limited to 10 paragraphs later. Many customers confuse premiums with deductibles. Next, would you like me to explain the difference for you? The room sat in stunned silence for a moment. The project lead turned pale, not understanding how it could have missed so badly on such an obvious question. The CEO, still processing his shock, declared, six months of work and $5 million? I don't understand. It's better than I ever could have imagined. How soon can we go live?

Raymond Moss: My gosh. it all depends on the audience, does it not? That's amazing.

Ben Curtis: And the perspective.

Raymond Moss: And the pr yeah, exactly.

Ben Curtis: Yeah, there was one other meeting somebody told me about where HR meets with all the employees. They're about to roll out AI within the agency. And the

Raymond Moss: Yeah.

Ben Curtis: HR team says, AI is not here to replace you. It's here to make you more productive. The main thing you need to learn to be successful is simply how to talk to it. Important aspect of AI. So they say, just imagine that you're supervising someone who's incredibly confident, occasionally brilliant. but also makes a lot of stuff up and can be completely wrong in a very self-assured way. One employee raises his hand says, I'm confused. Are we still talking about AI or is the management team gonna start working for us now?

Raymond Moss: That's so great.

Ben Curtis: You

Raymond Moss: Yeah, there's there's so much there is so much uneasiness about what the heck is going on with AI. Probably in every industry, but I know for sure in the trucking industry, because it has so many places that it can touch and that it's starting that it is starting to touch. And we're a couple of actual years into this AI revolution. yeah. How is AI touching your workflow? Is that what we're talking about today?

Ben Curtis: Yeah, so why don't you read the intro and we'll get into it.

Raymond Moss: Yes, so this is the podcast that accompanies the DOT analysis platform where we discuss real time trends, the truck insurance market, and explore answers to the questions you are not allowed to ask. I'm Ray.

Ben Curtis: And I'm Ben, and you're in the inside lane.

Raymond Moss: All right, so AI. AI in truck

Ben Curtis: Yeah.

Raymond Moss: Insurance.

Ben Curtis: Yeah, and even just in general, mean, has there been a more transformative topic in the last one or two decades that you can think of than AI? mean, you'd

Raymond Moss: Well

Ben Curtis: Have to go back to maybe smartphones. I mean, what is the last?

Raymond Moss: Yeah. The the last thing I can think of is when Steve Jobs came out and he said that he had the internet in his pocket and everyone's like, What? What do you mean? I I distinctly remember that time. And that really did change stuff. I think this is a very similar moment, for sure.

Ben Curtis: Yeah, yeah, it feels like it. You know, nobody knows quite how to understand it, seems like. And yet everybody has this sense that they need to figure it out. At least that's the increasing sense that I get.

Raymond Moss: Well, yeah, and a lot of people have a fear around it because things you don't understand naturally are y it's easier to be afraid of things you don't understand. And I've you know, experienced that in, you know, places in life where I don't understand something. It's like, wow, that's terrifying But then you get to a point of being able to understand it and it is less frightening.

Ben Curtis: Of course, yeah. You know, I think of it a little bit like back in grade school when you're first learning math. And of course you have to show your work and write all of your equations out. And then as you progress through school, you get into algebra and calculus and more complicated mathematics. And it shifts over from learning how to do the the algebra or the mathematics yourself and the class actually becomes understanding how to use a calculator or especially a graphing calculator and it becomes.

Raymond Moss: Yeah. Well, because you you know the underlying concepts at that point, right?

Ben Curtis: Of course. But now what you realize at that point is if I'm stuck solely in that, I'm never going to make it. To really do well here, I'm going to have to understand how to leverage the tools at my disposal to actually

Raymond Moss: Right.

Ben Curtis: Be able to progress. it feels like that is the point that we are at with AI, where people are, I think that's where the desperation is coming in, where people are realizing if I don't figure out how to utilize this, I'm going to get left behind. because it is such a game changer for the people who are really implementing it well.

Raymond Moss: Yeah, I think that I think that's very true because, you know, it it's I'm sure it's different for different industries, but and every industry has this component. But speed is a massive part of the game in truck insurance because speed to market, speed to communicate, speed to communicate accurately,

Ben Curtis: Mm-hmm.

Raymond Moss: Is massive. And that is one of the big things that AI does is it turbocharges the speed with which everything happens. And

Ben Curtis: Yeah, so before we get too far down the road with AI implementation specifically, there are some legitimate concerns and we at least want to hit these before we just get lost in the AI discussion. One big one is data security. And I know there's been

Raymond Moss: Yeah.

Ben Curtis: Some articles and studies out there, but why don't you talk a little bit about what the impact of or the risk factors of data security related to these models and training and feeding them data.

Raymond Moss: Yeah. that's a pretty big risk because now you're taking these third party tools, whether it's Chat GPT with OpenAI or maybe it's anthropic and claw or it's Google with Gemini and, you know, a thousand other different variations that are out there. I mean, it's like a new AI model or variance on a model. It gets spun up every day. But you start using those things and you're sending your data into where. And that is a really, really serious thing to consider. You know, when when AI hit the scene, there was the initial excitement of, my gosh, this this can process data super fast. But then pretty quickly people realized, wait a minute, where am I sending this data? I'm I'm reminded of a a story that came out, I think it was in twenty twenty three. I looked up I I went and looked for the article. Yeah, twenty twenty twenty three, where a Samsung employee actually took some of their proprietary code that they had been working on and uploaded it into the free model of ChatGPT, you know, a couple of years ago. Yeah.

Ben Curtis: Boy.

Raymond Moss: And it basically leaked because what a lot of people n I think a lot of people now probably realize this, but it wasn't very prevalently known at first, which is when you're using one of the free products from any of these AI companies, it's not free

Ben Curtis: Mm-hmm.

Raymond Moss: For you to use it. So the question is, okay, how are they making money? Well, they're you're becoming part of the product because now the data you feed into it. By default, they are training on it. And by default, they are using that data. And so if you're putting something into those tools that don't have the right security considerations, especially if you're in the financial services industry, like we are with truck insurance, right? You have access to a lot of private data.

Ben Curtis: Mm-hmm.

Raymond Moss: If you're feeding that into open models that are going to be trained on, like that is pretty serious. So I don't imagine that Samsung engineer still works at Samsung after he did that. But I don't know. So th there was there was

Ben Curtis: Yeah, likely not.

Raymond Moss: I think it was that incident and just around the time I remember seeing and you can go back and you can find the evidence of this where JP Morgan Chase and Bank of America and Goldman Sachs and Wells Fargo and you know a bunch of the big banks, right, basically

Ben Curtis: Mm-hmm.

Raymond Moss: Put down a full stop, no AI usage, you know, by their employees because they realized they were feeding in financial data of customers into these open models. And they said, nope, can't can't be having that. You have to have the right guardrails around the security and the privacy because you're dealing with really important data in our industry specifically, financial data, driver data,

Ben Curtis: Mm-hmm.

Raymond Moss: Fleet data, all sorts of data sets that are really, really important that they stay private and don't get turned into a public data set somewhere.

Ben Curtis: Yeah, so that's actually, I think, a perfect transition to where we wanna go next, which is looking at, so we wanted to acknowledge the legitimate risk factors that

Raymond Moss: Yeah.

Ben Curtis: Exist, and yet at the same time, the compute power and the ability of the AI, like the most powerful calculator out there, is just not something you can stick in the drawer and ignore. You've gotta figure out how to use it.

Raymond Moss: Right. 'Cause other because otherwise you know, and I've s and I've started to see this with some people, th you get left behind. I mean, there's not many people on the road with their horse and buggy anymore. I mean, if you want to go to the grocery store, you need your car.

Ben Curtis: There's not.

Raymond Moss: You're gonna get left behind if you're still using the old tools.

Ben Curtis: Yeah, yep. So what we wanna spend some time on in the show today is looking at some effective ways to deploy AI. And we wanna, we're actually gonna test real time here some of the tools and actually see how effective they are at analyzing some of this data. So along with data security, one of the big topics that we want to consider and we want to share is this idea of integrating AI into your workflow.

Raymond Moss: Yeah. And that's super key.

Ben Curtis: As a lot of people that we've talked to here are talking about ways that they are seeking to integrate AI, a lot of times it involves taking a specific document or taking a specific data set, feeding it into an external AI, trying to get some analysis or information or answers out of it, and then trying to bring that back to their workflow.

Raymond Moss: Yeah.

Ben Curtis: That not only has some of the inherent risk factors with it with the data security, but also is just

Raymond Moss: Right.

Ben Curtis: A clunkier process. So we want to look at what does that in context AI integration right into your workflow, what does that look like? How effective can that be? And

Raymond Moss: And specifically in this niche industries workflows themselves. And one of the things you didn't mention, Ben, is that that clunkiness, not only is it inefficient oftentimes when you're trying to use some of these ad hoc AI tools, which you can get

Ben Curtis: Mm-hmm.

Raymond Moss: A little bit of efficiency out of that. That that's good. But it if it's not built into your existing workflow, one of the things I often see when I'm working with end users, and that's one of the things I love to do is you know, working with and observing real end user work, is you can inject error, human error, because you're trying to move data between these tools and your workflow. And that becomes

Ben Curtis: Of Yep.

Raymond Moss: A another issue.

Ben Curtis: Right. So why don't we, in each of the shows here, we typically use the DOT analysis platform. A lot of times we're looking at market intelligence or market stories, but there are some AI tools built into that workflow. So let's pull that up and let's look at some of the ways that truck insurance professionals are utilizing AI in their workflow to do some of these powerful tasks.

Raymond Moss: Right. I I I pulled up a couple different carriers here and I'm gonna pull up DMARC Inc. and I'm on the losses tab here. And I'm just gonna take one of these PDF files and we'll upload it. DOT analysis, this is a great tool. It starts processing it

Ben Curtis: Yeah, so just to give all of our listeners context too, because not everybody's watching this, you're just listening. So we're in the DOT report section of the DOT analysis site. And within that, we can look up the full report on a motor carrier. And there's some integrated AI tools. so like Ray just mentioned, we're on the losses tab in that tool,

Raymond Moss: Yep.

Ben Curtis: Which has the ability then has an AI engine built into it that allows you to bring in loss runs and that AI is going to read the loss run and analyze the data. And so we're Watching it analyze the data, we're gonna test it on a couple files here and see how it does at reading those losses and understanding them from that truck insurance perspective.

Raymond Moss: Right. And I'll and I'll just point out again too, helpful for those who didn't just see that, who are listening on maybe Apple or Spotify, but in the time that you know Ben was talking there, it read that entire nine page PDF of losses and pulled out fifty five losses from that PDF document and puts that discrete data on the page, which is incredible

Ben Curtis: Yeah.

Raymond Moss: That it does that.

Ben Curtis: Yeah. So why don't you open the source file and

Raymond Moss: Yeah.

Ben Curtis: Let's just look at what it just read and then we'll look at the analysis.

Raymond Moss: Yeah, so here's the source file. And you can see I did blank out the last names of drivers that are on here because one of the cool features is it pulls out driver data, equipment data, it pulls out the cover like everything it turns all of this into discrete data exactly as how you'd want it. And by the way, we talked about security at the beginning of this episode for AI products. The data that you're feeding into this AI model is not going to be used to retrain. you know, public models. It's not going to be used like that. So it is a secure and confidential way to leverage AI in your workflow, which is really, really important.

Ben Curtis: Yeah, that's great.

Raymond Moss: So, okay.

Ben Curtis: All right, so what are we looking at on this loss run specifically?

Raymond Moss: So here's what's really fascinating about this last run. I had I'm gonna go to page five on here, and I just wanna point something out. And I'll kind of say it as I go for the people that are just doing the audio version of the podcast here. But if you if you have a chance, this is a great episode to watch because being able to see this is pretty, pretty amazing. So here's an interesting claim. this is a claim. I'll just read the description of it here. The description is as per insured, driver T-bone C V school bus. Well, number one, that sounds very sad, you know, of you know, of a accident. And being in the truck insurance industry, you deal with stuff like that all the time. so that's a terrible thing to have happen. But what I want to point out here is we have a claimant, grandpa's busco, and we get a coverage. Now, this is so common because of what I what I'm about to describe is so common and it's that you get someone that's Not necessarily even new, but someone who's only been in this industry for, you know, a marginable amount of time, a year, two years, it's very difficult sometimes to attain the skill of reading loss runs accurately. And I'm looking at this and it says grandpa's busco. Well, this this thing was T-boned. And what does it say on the coverage line there? It says PD. Well, I can't tell you how many people will look at that and go, Well, golly, they're replacing the bus. And it was T-boned, obviously it's physical damage. So they look at that and they think that's a physical damage claim. Now they put that, here's what it is: $35,000, almost $36,000 worth of loss. Actually, the gross incurred is, you know, in that $36,000 range. That's a pretty big dollar amount to have put into the wrong

Ben Curtis: Mm-hmm.

Raymond Moss: Coverage type. Okay. So let me let me just think about this for a second.

Ben Curtis: You

Raymond Moss: Is this physical damage? Or is this not physical damage? Because I see the PD and my gut wants to say, well, grandpa's bus, that's physical damage. We're gonna replace the bus. Well here's the reality.

Ben Curtis: Mm-hmm.

Raymond Moss: I know, because of my industry knowledge, and many of you know in this audience, that this is a loss run for the motor carrier DMARC Inc. Well, think about it this way. Grandpa's busco, that's a third party. Physical damage is Think about how first party and third party coverages work. If this was physical

Ben Curtis: Mm-hmm.

Raymond Moss: Damage, it would have to be a first party claim. This exactly,

Ben Curtis: Yeah, DMARC would have to be the claimant.

Raymond Moss: DMARC would have to be the claimant, but that's not the case. You have a third party as the claimant. So I know that that PD does not stand for auto liability or does not stand for physical damage. It stands for auto liability. That's the property damage of the auto liability responding to grandpa's busco. So I know that. Let's see if the AI knows that. All right, so let's switch back over to DOT analysis and let's see if the AI got that right. So I'm gonna open up my list of losses here, and I'm gonna find that loss from twelve five. And here it is, twelve five, and here's look at that, thirty six thousand dollars. AL, it absolutely got the fact that the coverage was auto liability. Because what does it have here? It's got that grandpa's bus co. as the claimant. So it knows that it is a third party making the claim because obviously it's the the loss run for the motor carrier. So having that sort of knowledge that is fine tuned and I will note standardized because any agency, any underwriting group is going to have multiple people in on their team that's going to be reading these loss runs. And let's say you have 10 people that, you know, that do this. If just one of them doesn't understand this, now you have a non standardized process in your organization. And that can breed some big problems, for sure.

Ben Curtis: Yeah, now I do see there's an update button there, so it looks like I could make edits to this screen if I needed to.

Raymond Moss: Yeah, you can.

Ben Curtis: And I don't think what we're saying with a tool like this and what's important with any AI analysis is that it's gonna do the starting point in the heavy lifting.

Raymond Moss: Right.

Ben Curtis: Like this just did a ton of data entry in a matter of seconds

Raymond Moss: Yep. Right.

Ben Curtis: And did that analysis, but if I am responsible for analyzing this loss run, I'm still gonna be comparing some of these lines to the source file. I'm still gonna be double checking some of its work and it looks like I've got the ability to generate this summary in the way that I need to see it.

Raymond Moss: Yeah, exactly. So there's two use cases for that. You can, of course, there's always a time when you want to allow AI to do the heavy lifting on all that data entry, but you want to double check, you know, certain aspects of it and j and just eyeball it and make sure things are good. Well you can also do is you can add data to it. So for example, on this loss run here, you'll notice that inside this claim, it didn't have the VIN that was associated with that claim. But I might wanna add that here. So I can add the VIN, I can add the equipment that was being used when that occurred.

Ben Curtis: Sure.

Raymond Moss: And so I can enrich this data then with, you know, anecdotal data that I need.

Ben Curtis: Yeah, because of your firsthand knowledge of the account if you're working on this with your customer. Yeah, that's great.

Raymond Moss: Correct. Yeah. Yeah. Exactly. So if I want to create a complete picture of this, I can do that, which is really, really nice.

Ben Curtis: Yeah, so another thing that's common, of course, is you're gonna get multiple loss runs.

Raymond Moss: Right.

Ben Curtis: So what happens if I've got three or four loss runs for this account?

Raymond Moss: So yeah, good question. let me shrink the spec down and then I'm gonna take a loss run from Acuity and I'm just gonna drop it on here and it's gonna start analyzing that. And then I have another one here from Berkeley and and I'm gonna drop that on that. So it's really nice to be able to process multiple of these files simultaneously, and then boom, they're done. It's you know, on one of them there was another twelve losses, on another one there was nine losses, and now I have all the losses by year. on here, which is really, really cool.

Ben Curtis: Yep. All right, now if I'm thinking about this like a CSR or an underwriter though, and I just dropped all these files in, let's say that I wanna go to that acuity loss run and I just wanna look at that one because right now I've got a lot of data all merged together. How would I do that?

Raymond Moss: Right. So to one thing I want to point out to you here is that if you if you click on this, you open up the raw file, right? But maybe you want to see the analysis of just the data that was extracted. Well, I can filter that and we put all

Ben Curtis: Right.

Raymond Moss: Sorts of filters in here that you would want to filter by. So I click on insurer and I click acuity and now I'm just looking over here at the losses that were on that acuity loss run, which is just so nice to be able to see that. You can see them by year and cleans it up and

Ben Curtis: Yeah, that's amazing.

Raymond Moss: Makes it really easy to focus on the data that you care about.

Ben Curtis: And I see that coverage button too, so does it also let me filter by coverage type?

Raymond Moss: It does. It'll let you filter by coverage type. So I can say, you know what, I just want to see, for example, the auto liability claims that were on this acuity loss run. And now I'm looking at the three claims, or the three losses rather, that that were on here with acuity. It's super nice to be able to do that.

Ben Curtis: So then I assume if I clear that acuity filter, then it's gonna show me auto liability losses from all three loss runs that are there.

Raymond Moss: Absolutely. Yep. You can mix and match all the filters. Really

Ben Curtis: Yeah, that's great.

Raymond Moss: Nice to be able to do that. And one of the things that's one of the reasons that that's so nice is because yes, this this is three loss runs, and you might have three loss runs easily on most f small fleets, but I've seen it where you can get nine, twelve, fifteen, twenty loss run files for a large fleet. And my goodness, that is overwhelming to be able to do. That might

Ben Curtis: Yeah, no doubt.

Raymond Moss: Be the weak that a CSR has to spend organizing all that data in an understandable way to present a coherent loss picture, you know? Really nice to be able to save all that time.

Ben Curtis: Yeah. All right, so we spent a lot of time on loss runs, but that probably is one of the heaviest, most time consuming tasks. But

Raymond Moss: Right.

Ben Curtis: Before we wrap things up here, let's look at a few of the other key documents that are common within the industry and that AI can really move the needle on here. Another big one that I'm thinking of is IFTAs.

Raymond Moss: Yeah, absolutely.

Ben Curtis: Do we have an account we can look at and see what IFTA data?

Raymond Moss: Yeah, I I I prepped one here for you so we could do that pretty quick. I got

Ben Curtis: Okay.

Raymond Moss: This Anthony PL, this motor carrier that's out of Iowa. And what you what we can do, and everyone's familiar with those IFTA data or that those IFTA documents, right? They come in all sorts of different formats. Th

Ben Curtis: Mm-hmm.

Raymond Moss: Oftentimes they have that watermark in the background and it's sometimes it's multiple pages. It's gonna have

Ben Curtis: Right.

Raymond Moss: The number of miles per state. it's that, you know, it's that tax document. And it's really time consuming to be o to jot all that down and try to get a picture of where that motor carrier is actually traveling to because that's the whole point, right? You're using the data on there.

Ben Curtis: Right.

Raymond Moss: So

Ben Curtis: Well, not to mention that writing that down still only gives it to you in number format.

Raymond Moss: Right.

Ben Curtis: And really this is data that's best visualized.

Raymond Moss: Right, it is. It's really nice to visualize it.

Ben Curtis: That's what I'm seeing on the screen right now is we've got a map and it's highlighted. So first of all, I see that you're on the mileage tab now of the DOT report and I see a big map view. So tell us in a bunch of dots on there, what is on the map view? What are we displaying here?

Raymond Moss: So here's what we're displaying. So I took four PDFs and I just dropped them onto this page here. And the AI engine within just a few seconds read those PDF documents. It was the miles for all those IFTAs, or for this motor carrier, for each one of those quarters, and it just plotted it on this map. And then it shaded the states according to the percentage of miles in each state, and it gave me it gave me the

Ben Curtis: Yeah, wow.

Raymond Moss: Percent. And I can check this out. I can toggle so I can see either per the percentage percentages of miles in each state or the number of miles in each state. So I can see absolutely a visualization of where a motor carrier is going.

Ben Curtis: Yeah. So a couple other things I want to point out here then too, and also for those only listening, we've got yellow and red dots on the map, which reflects the inspections

Raymond Moss: Right.

Ben Curtis: And the crashes. So that lets us visualize that data overlaid with their mileage data, which is really insightful for a variety of reasons. and then

Raymond Moss: Yeah. Really nice to be able to to to layer on that data. And and it's a good comparison too, because check this out. With this example, they have inspections that are up in the northwest of the country. So Washington, Oregon, you know, they have Idaho here, and they have inspections there, but they don't have any miles reported there. So there's something fishy going on here. So if you're the underwriter, you absolutely want to know that because you are you're dealing with something that might not be up to snuff.

Ben Curtis: Or if you're the agent getting ready to submit this to the underwriter, you want to know that before you make the submission.

Raymond Moss: Exactly, 'cause you know you know the underwriter's gonna find it and you need to get ahead of that issue.

Ben Curtis: So the other thing that we've had a lot of success with our clients doing is being able to walk through the development of these miles. So I notice on those bar

Raymond Moss: Yeah.

Ben Curtis: Graphs there, they're all toggled on, but you can toggle on one quarter at a time. And then as you toggle on each quarter, you get this progressive look at how their miles are changing and how maybe their routes or the states that they're in are changing over time. It's hard to say. what exactly is going to be meaningful data. But the ability to visualize this allows you to understand potentially meaningful trends in that data, Like that's why the visualization is so powerful. You can't always predict what the nugget of value is going to be, but now all of a sudden you identify, hey, they've drastically reduced or eliminated mileage in an area that the underwriter is less excited about. So just being able to visualize and correlate all of that data oftentimes generates really unexpected but very powerful insights.

Raymond Moss: Yeah, absolutely. Yeah, really great to be able to do that. Visualizing big data helps you make use of that big data. And if there's one thing that this industry is drowning in, it's data. There's so much of it out there. How do you make use of it? And this is a amazing way to make use of that data.

Ben Curtis: All right, so we gotta move quick here, because this has been a lot of time. But let's look at the equipment tab quickly and see what we can do with equipment list. This is another big one. Not quite as scary of a document as a loss run, but still a lot of numbers and a lot of data. How can AI within this workflow take that public data, take private data, and make sense of it and make it meaningful quickly?

Raymond Moss: So let me give you just the quick overview of this. I'm looking at on the equipment tab, in the DOT report, I can see that this motor carrier has nineteen power units and fifteen trailers. Now I see those VINs because that's what's available publicly in the data, right? So they've had inspections on all those. Based on inspections, yeah. But of course, if I'm gonna submit

Ben Curtis: Based on inspections. Okay.

Raymond Moss: This to the underwriter, I absolutely need to get a real equipment list from the motor carrier. So here's an example. I'm gonna drop my my list of equipment into the do report here. And what the what DOT analysis does is it reads that equipment list off the PDF or off that Excel file, whatever it is. It's great when it reads it off the PDF. And here's what happens. Now I know, okay, they don't have just 19 power units. They actually have 21 power units. And they actually have not 15 trailers, they actually have 20 trailers. But now check out the comparison that happens automatically that the AI does. The AI, if you can't see it on the screen because you're listening on an Apple or Spotify It's putting a yellow triangle by two of the VIN numbers and the and it's telling me this message. It's saying this VIN was in it was inspected on 424 26, but does not appear on the imported equipment data. So it's saying that for two of the power units and two of the trailers, it's basically telling me, hey buddy, hold on a second. This this motor carrier has reported equipment to you, but didn't report two of the power units that we absolutely know they're driving because one of them they were driving what is that January, February, April, and one is in June. So they were they just got inspected with this VIN, but it wasn't on their equipment list. An underwriter

Ben Curtis: Hmm.

Raymond Moss: Is absolutely going to ask about that. And if you're an underwriter and you're using this, you're absolutely going to ask the agent what's going on with that because you either need a bill of sale to show,

Ben Curtis: Right.

Raymond Moss: Yep, we sold that piece of equipment, you know, yesterday, or you need to understand the story behind why is this you know, not on your submission list. Are you trying to hide qu hide equipment? And it told me that in five seconds. You know, so again, speed.

Ben Curtis: Right.

Raymond Moss: This is all about speed. To be able to find those insights is incredible.

Ben Curtis: Yeah, absolutely. Okay, this is, we've looked at loss runs, we looked at IFTAs, we looked at equipment lists, those are some of the largest data heavy documents that get processed as part of this workflow. We looked at all ways to take that data now, and this is all enhancing and building out one comprehensive profile within the DOT report where all that information lives in one place. Like you said, we're not taking data and extracting

Raymond Moss: Right.

Ben Curtis: It. and taking it over to an AI somewhere else, trying to bring it back in or vice versa. This keeps it all in one

Raymond Moss: Correct.

Ben Curtis: Place where it all works synergistically together. I can see it all in one place and we're correlating all that data with each other.

Raymond Moss: Right.

Ben Curtis: There's a couple other AI integrations here that are gonna look at that publicly available data. Let's look at Cargo quickly and then.

Raymond Moss: Yeah, this is so cool. So let me let me pop back over to this one and I'm gonna hit cargo and I just wanna point this out. So everybody knows that there's cargo listed on an MCS one fifty, right? And that is what a motor carrier is telling you that they're doing. But buried inside this data is information about their cargo that if you went through and spent hours, you could find out, or you can build an AI model that will do that deep research for you. So here's what we do it in DOT analysis on the DOT report is there is an AI agent that looks at all the different shippers. So all the different actual loads that this motor carrier was hauling down the road and it does an analysis and a deep research of all those shippers.

Ben Curtis: And that's those shippers identified by inspections, right?

Raymond Moss: Correct. Those shippers that are identified by inspections. And then it does a deep research analysis of those, the quantity with which they are being inspected, and it will tell you exactly what is being hauled. So in this case, meat and poultry, dairy and ice cream, beverages, retail and distribution centers, ag and food processing, just absolutely phenomenal analysis of cargo that, you know, is would take hours and hours of worth of research in order

Ben Curtis: Yeah.

Raymond Moss: To sift through all of the shippers that are on a motor carrier's history.

Ben Curtis: Yeah. Well, that's amazing. And I see all the little blue bubbles of this listing out

Raymond Moss: Right.

Ben Curtis: The shippers under each category. So even if you have deep insight with this customer, they have an explanation for why they're not carrying something and what that means. It allows you to quickly identify what that category is being derived by. you can still take your intimate knowledge with a customer and still build out the narrative that makes sense. Still work with the underwriter. to get the information you need. This just gives such a automatic and quick ability to analyze data in a totally unique way and give you a lot of depth of insight into a motor carrier that would otherwise just take a tremendous amount of work to try and un.

Raymond Moss: Right, exactly. Cause in this case, you know, that MCS one one fifty listed cargo says nothing about meat and poultry. But you know what? This motor carrier's largest item, meat and poultry and then dairy and ice cream. You you can see that in the data, which is just phenomenal, that you can do it that fast.

Ben Curtis: Okay, so one last thing, within the DOT report, there's all of that public inspection data, all the violations. Let's look at the AI tool here that analyzes all of that and displays automatically some trends for us and look at how powerful that is.

Raymond Moss: Yeah, so this is pretty cool. So if I go to the summary tab on a DOT report and I scroll down to the bottom here, I can see both the positive trends and the negative trends. And the reason that we designed it to pull out both is because, yeah, everyone is always talking to trucking to the trucking companies about the negative things. You're speeding too much or you're lying on your logs, or you're you know, you got vehicle maintenance issues. But what about the fact that motor carriers and trucking companies actually do great stuff. You know, they are improving. They are training

Ben Curtis: Mm-hmm.

Raymond Moss: On safety. They're doing good things too. So in that data, we identify positive safety trends. So here's an example. Driver, driver activity violations decreased by 13% over the last 12 months. You can you can tell that trucking company, good job. You're you're doing better by 13% over the last

Ben Curtis: Yeah.

Raymond Moss: Year. Here's another one. Hours related violations. That's decreased by thirty nine percent over the last twelve months. Nice job, right? Wheels violations. You're doing better

Ben Curtis: Yeah.

Raymond Moss: On your wheels. Having an AI agent, having a na you know, instantaneous analysis where you can point out something good about the prospect or about the client, about that motor carrier, is so helpful to them because I can't tell you how many times I've heard this where motor carriers say, you know, the insurance people, they're always looking at the negative things about me. They're they're always pointing out the bad stuff. Well, maybe you You wanna point out the good stuff. DOT analysis pulls the good stuff, surfaces it to you, and then lets you leverage that so that you can build relationship with that motor carrier. That's pretty cool.

Ben Curtis: Yeah, absolutely. So obviously we're pretty excited about the DOT analysis platform specifically, but I think what we want to leave all of our listeners with is AI is a very powerful tool. You've got to figure out ways to start using it effectively within your business. Whatever role you play here, agent, producer, CSR, underwriter,

Raymond Moss: Yep.

Ben Curtis: You've got to be leveraging this tool. Now you need to do it safely and you need to do it effectively. This is one of the best platforms that we're aware of that brings so many of those tools into one place and integrates them into the same workflow. But you've got to find a way to be able to do that and leverage the horsepower of this AI, especially for these, I'll call them mundane tasks. Not that they're not important, but there's just so much menial work involved in analyzing that data.

Raymond Moss: Right.

Ben Curtis: And so being able to utilize a tool that's going to turn that into valuable insights. that really move the needle is super important.

Raymond Moss: Yeah, absolutely.

Ben Curtis: Okay, well, that was a lot of time spent on AI, but we've spent so much time talking about this within the industry and within our customers. So we wanted to do a deep dive here and unpack some of the capabilities and talk about the exciting ways that it's being used and how it's helping transform people's businesses and improve the lives, especially of those people who spend so much of their time doing that hard work of the analysis and understand how you can really change. your life and change your experience by harnessing the power of AI to do a lot of that work for you. So hopefully this has been helpful. If you've got new implementations or other unique insights into this, we'd love to hear your feedback. So let us know how you're using this or the questions you have and maybe we'll cover that again on an episode soon.

Raymond Moss: All right, we'll see everybody later.