Demos & Shared Insights - Linqura | Sixfold

Immerse yourself in an engaging series of technology-focused presentations, unveiling an array of innovative products and services specifically tailored to enhance your workflows.


Transcription:

Arlan Mocha (00:08):
Good Afternoon everyone. I think we're ready to start the afternoon sessions for the demos. My name is Arlan Mocha. I wear two hats. I work for HSB on my day job. And then I also am part of insurtech Hartford, the organization that set up insurtechs. So you guys ready to immerse yourself with two engaging, innovative companies that are focused on enhancing your workflows. And I hope that everybody's caffeinated. The bar is back up. I see a lot of sugar and ice cream, so hopefully we'll have everyone's undivided attentions for these two great innovative companies that are joining us. Let's see. So I'd like to welcome on stage Brad Stender, who's the VP of Product Development at Linqura. Linqura they use advanced AI techniques to make high quality decisions and with more, Brett is going to tell you. Alright,

Brad Stender (01:00):
Thanks Arlan. Appreciate you.

(01:05):
So thanks Arlan, appreciate the introduction. I guess I'll turn and face most of the crowd. So I'm with Linqura. We are an AI company that has collectively over a hundred years of experience and insurance across all different sections of the insurance lifecycle. And we've also got 40 years, excuse me, of AI expertise and experience dating all the way back to Watson Labs. So what you guys are going to see today is Sybill, which is our agent copilot for insurance. So a good way to think about Sybill is a link to as a window into our link 1.0 platform. So we have a larger AI platform that is trained and built on insurance data. And so we're able to understand insurance transactions and predict insurance outcomes. Why is that important? Well, mostly if you're using an open AI or one of those more generic models that's not trained specific to insurance, you're going to have essentially a vacuum cleaner across the internet just pulling word out together.

(02:25):
You have no idea if it's accurate or not. So we have a major accuracy advantage there. So how this came together is we worked with an exclusive partner with Rough Notes, who is one of the oldest publications in insurance, and they have collected 750 different versions, or I'm sorry, different risk classes that we now have, coverage recommendations, exposures, mitigating factors, and all the different things that you need to know in order to write each particular risk. And so what we've done is we took that rough notes content, we married it with 44 million US business firmographic records, and we've essentially created insurance profiles for 40 million different businesses in the United States, and we're continuing to grow that every week. So ultimately what that means is now the agent at the point of sale goes into each situation, whether it's a new customer, whether it's a renewal scenario with full understanding of the business that they're looking to write and the insurance needs of that business.

(03:40):
They do the size and scope and they have the insurance needs. And so we see it as a real growth play, retention play and EO opportunity to be at the front and center with the insurance agents. So with all that, I'm going to jump right into the demo here. So the first thing I'm going to do is select, I have a new prospect and so Sybill is going to prompt me immediately for the business name in the city. And so I'm going to go head here. We're down here in south Florida. I'm going to pick a barbershop down in South Florida. And so here we did a quick lookup of a record and you see we got the business name, we got the address, we got the Nate Star description, we found the right business right now from here, if I'm a new agent, there's certain things that I need in order to write any sort of business. So I might ask, what is the workers' comp code? And so Sybill's going to return an answer here, should detail the workers' comp code. And so the next question I'm going to ask is, what is the GL code and why is this important?

(05:08):
And so we should get the GL code back with an explanation of why GL codes are important. So we got the GL code back in context of Adam's barbershop. The code is 1 0 1 1 3. And so now the agent has the ISO code that they need to write the barbershop. So once I have the necessary codes, I've got Nate's code, I've got worker's comp code, I've got GL code, well then I might want to know the risks or I'm sorry, the size and scope of the business. So I'm going to ask, tell me the sales revenue and number of employees. And so this is where we married those business firmographic records. And this is really going to start to come together. So the agent's going to have the size and scope of the business. So this should return the revenue amount, the number of employees, that's all real data, that's not AI generated data. And that's something we're working with best in class of service providers on in order to continue to enhance our dataset in the insurance profiles. So now I understand I found my business, I understand my codes, I need, I know the size and scope of the business. Let's dive into some risk. So I'm going to ask a very generic prompt, sort of a higher level, what are the risks I need to know?

(06:37):
So because I asked a bit of a higher level prompt, this should return a one through eight list of different kinds of exposures that we need to think about in context of Adam's barbershop. So we get a list of exposures one through eight and we get one line descriptions. Now you'll notice I'm able to sort of, before the answer appears, tell you guys what the answer's going to sort look like. That's because we use a little bit of a different method from if you were going to use something like a ChatGPT where everything is all retrieval on augment degeneration, and it's sort of a nets best word filter. If you ask the same question three different times, you're going to get three different answers. We do something a little bit different so we get some answer consistency, which you're sort of seeing as I'm able to talk through to the answers before they appear. So now I've got the risk, right and sort of a high level risk, but I want to dive into it. So I'm a new agent. Tell me more about the product liability exposure.

(07:43):
I have a little bit of a typo in there too, just to see how we handle it. And right now we should get a breakdown of product liability exposure. It should have a score. So the product liability exposure is typically moderate, right? We get a list of factors in which if we had that information, it would change the exposure level. So if they were to sell, for example, their own proprietary products, you'd see an increase in the exposure. And then we'd get some details around how to manage the risk. And so I can do that for any number of exposures. I could do that for workers' comp, I could do that for general liability or sorry for property. But the next thing I want to do is sort of detail going to causes of loss. So I can actually target individual causes of loss. If the agent knows, for example, that there's been claims around slips and falls, we can actually ask, how do I reduce the risk of slips and falls? And again, notice I haven't had to mention the business again, right? Sybill understands that the whole context is in the conversation of Adam's barbershop. We don't need to continually mention that in order to retrieve contextual results. So

(09:07):
We should get an answer here. It might, we're going to go here. So

(09:24):
When I asked how can I reduce the risk of slips and falls, we generally we're going to supply a list of things you can do, right? Ensure you have good lighting, make sure you have the stock on the shelves is moved, right? So it's not on the floors. Maintain adequate flooring, clearly mark your steps, keep your parking lots clear, just a list of things you can do to reduce the risk. And then where I think this really becomes valuable for the agents is the recommended coverages. So we're able to prompt Sybill for the recommended coverages. And what you see here is a list of coverages with sort of a one line description. So that comes from that exclusive data we've got with rough notes, and we're actually in the middle of building out a mapping to the actual cause of loss for the business. So we'll be able to take this list that you've got of 13 that are already recommended coverages and really narrow that down to the top five or six that you definitely need in context of your business.

(10:24):
So that's something that we're actively working on that we're really excited about. And then just, I'm sure we're on time here, but if I'm a brand new agent, I just picked an obscure coverage type, right? Valuable papers and records, I might have no idea what that is. So I can prompt for valuable papers and records and we'll tell you in context of that coverage, what that coverage is, why it's important for the business, and then ultimately we can ask questions about bombs or CPPs or any sort of packaging. How do we package this as an agent just so they know how to write the business. And so we've got all those details as well, what coverage is applied to which policies and how you're supposed to bundle these things together. So I'm close to the end of time here, but essentially we're starting with this education use space, but we have this larger AI platform that is API enabled and is models that are built on insurance data that we can plug into any place in any core system in the insurance lifecycle. And we can address underwriting issues, we can address placement issues, market appetite. So there's a lot of big exciting things coming. If you're interested in hearing more, come find me. I had love to talk to you or come find my lookalike over there. So appreciate you all. Thank you very much.

Arlan Mocha (11:57):
Thank you, Brad. Definitely helping out with the agency experience there and streamlining that workflow for sure. A round of applause again. Alright, next I would like to welcome Gregg Tourville, who Heads up Design at Sixfold. Actually, right before this I was chatting with Greg and I wanted to ask him, Hey, can you tell me a little interesting fact about yourself? Have you done parachuting paragliding or something? And he's like, no, I live, breathe and eat Al's. And he said, that's what ChatGPT told me to say. So in order, here you go, Brad, it's up to you.

Gregg Tourville (12:32):
I'm good. Thank you. Thank you. Yeah, no, actually I eat, breathe underwriting, right? Yeah, there we go. Yeah. Hey, I'm Greg. I am the head of design at Sixfold AI and we are the first generative AI tool exclusively for underwriting. And I have a feeling you've probably heard AI roughly 67 times in the last, let's call it hour. So I'm not going to focus on that. I'm going to focus in my demo on the outcomes that we can bring for carriers, MGAs and reinsurers, and the fun, the joy we can bring back to underwriting itself. So with that in mind, how do we do it? Our superpower is taking in your appetite, understanding what you want underwrite and what you don't want to underwrite in really simple natural language. And then taking in all the risks that are coming in, all of the submissions and analyzing it based on how you would underwrite it, how your underwriters would look at it.

(13:42):
And then returning a score and returning all the information your underwriters need in order to properly do that underwriting, to speed them up and to make sure that they are paying attention to the risks that you want to underwrite the most. So let's take a look at it and we're going to start at the most exciting part of any app, which is the settings page, of course. So here we have a fake underwriting guidelines that were set up and we can ingest these guidelines in natural language. We see that we can say these types of companies, these classes of companies, we either, we really want to underwrite, this matches appetite, we don't want to underwrite them at all, we're going to disqualify it or we're wary. Everything else has to align for us to underwrite it. Similarly, we can look at particular aspects of a case of a risk, such as in the cyber case, is there a CISO that obviously is a good sign that matches appetite.

(14:52):
On the other hand, if they've had incidents, we want to know that. We want to know what those incidents were that's going to reduce that match. And finally, I mean more of the underwriting really happens in digging into a case and understanding the details. So we can ask any questions you would want your underwriters to ask that you would say in your underwriting guidelines, you need to look at these things. We can set those up and pre-configure that so that they're always looking at those things. Alright, so let's see what an underwriter would see. Alright, here on this page, not much going on. But what you see and what's most important is the triage aspect. We know that your underwriters are going to get every hour, 10, 50, a hundred different submissions coming in. And where do you want them to spend their time, right? You want them to spend their time on the things that are going to bind, that are going to increase your gross written premium. And you want them to ignore the things that are clearly disqualified, like these zeros and ones we've got. So let's take a look at one of these 23 and me.

(16:05):
So I've run this case already. It takes us about 30 seconds to run a case, but all we need to start a case are the name of the company and an address or not even address, really just a state. And then any of the information that you would give an underwriter, any of those documents, any whatever format they're coming in, any of that, JSON, any of the PDFs, whatever else, give it to us and we're going to analyze it. And what we see is we're focusing the underwriter's attention on the things that are most important, the things that have matched or not matched your appetite. So here we see they have an intrusion detection system, that's great, good sign. However, they have had a cyber incident and we'll come, we'll see. It looks like they had a phishing attack. And we can dig into that more because we know we've asked the question, were there any cyber incidents?

(17:03):
Where did it happen? So we'll come down here, take a look, and we see that it happened a couple years ago and they did take steps to resolve it and to prevent it from happening again. We'll also see exactly where we pulled that information from. And I mean, I know this is crucial. I'm trying not to talk about ai, but I know there are trust issues with ai. And so it's really important that we're able to link directly to the page that we found this information in the documentation or the website or wherever else we are going to, or the third party information that we got this from on top of all that. 23 and me, it's a known entity, but a lot of his submissions are going to come in from companies that you don't know what they are. So we'll give you a description of that company and we'll also give you the NAICS, the SIC, whatever other business classification you want. We've actually seem to have a best in industry matching for business class. So that's great. So that is the PNC side of the house. But because of the way we've set this up, you can set up lines to not just do, or you can set up any line of business that you want on the p and c side, but we also do life and disability. And the idea is similar, but we'll take a look real quick.

(18:23):
So to start kicking off a case, which I didn't show in PNC, all you got to do is come in here, we'll throw in an APS. And in this life and disability underwriting, it could be tens, hundreds, thousands of pages that are coming in. And that's just way too much time for an underwriter to properly go through. So we're going to look through that and pull out the most important pieces. And I think something really critical, well here, we'll just start this off, make up some stuff. There we go. And talking about the joy of underwriting. Look at that. Isn't that cute? While we wait for this to run, but I'll take us back to a case that's actually run already. Here we go. So while we've analyzed all these medical records, we don't want to turn a hundred pages into 15 pages of lists of medications and diagnoses and a visit to a doctor just because of a cough that's not useful.

(19:40):
That still is going to require the underwriter to run through a ton of information that's really just noise looking for the signal. We want to pull out that signal immediately. And that's why we, again, in this top section, we're going to call out all of the things that are concerning based on your appetite. And if we want to dig in more this leukemia that might be concerning for either life or disability. But we see it's been in remission for over 10 years at this point. So maybe that's not a concern anymore. And if we needed to dig in further, we can go in further. We have a full description of the most important parts of that medical history, and we can dig into each one of the potentially related diagnoses, medications, any lifestyle data such as substance use, all of that all at the fingertips of an underwriter and quickly summarized so they don't have to go and find it other places.

(20:51):
And with this, what we're really trying to do is we're trying to increase your gross written premium by making sure that underwriters are focused on the most viable risks on the thing that are most likely to bind. But doing that while increasing the accuracy of your underwriting, increasing the consistency of your underwriting, because we all know that each underwriter is going to do it a little bit differently and we want to do it in a transparent way, right? We want to show you where we're getting this information, whether it is from the documents, whether it's publicly available information or from a third party, and let your underwriters make the best decision so that you can grow your book. And with that, that's all I got. Thanks for listening, folks. I appreciate it.

Arlan Mocha (21:43):
Thank you, Greg. And with that concludes our round for the demos. Congratulations and thank you to Linqura and Sixfold for some great demos. I know we finished quite a little bit early, so that means more time for you guys to network and socialize before we have our last round of breakout sessions. And with that, you guys can head out to happy hour and do day two again tomorrow. Thank you. Have a great rest of the day.