How Modern Agencies are Using AI to Find and Keep More Clients


Chris Cheatham, Co-Founder & CEO, RiskGenius on the Connected Insurance Podcast presented by Agency Revolution

Chris Cheatham, Co-Founder & CEO, RiskGenius

How is Artificial Intelligence (AI) giving some modern agencies an edge over their competition?

Our guest Chris Cheatham is the co-founder & CEO of RiskGenius – a company that offers independent agencies an AI-based tool that takes the hard work out of finding the right policy for your customers. In this rapid-fire conversation, Chris and our host Michael Jans look to the fast-approaching future of the modern insurance agency, and how agents are using AI now:

  • How AI slashes the time it takes to compare, review, and analyze policies
  • How agents are using AI to outpace slower-moving competitors
  • How AI can identify gaps and help agents provide better protection

Whether you’re a personal or commercial lines agent, this is a glimpse into the Modern Age of Insurance you don’t want to miss. 

Presented by Agency Revolution, the Connected Insurance Podcast provides weekly opportunities for listeners to dive deep into the trends affecting insurance agents and brokers today and to gain proven strategies and tactics for agency growth. Our hosts facilitate thoughtful panels and 1:1 conversations with a variety of prominent thought leaders, with a focus on how to streamline and drive operational efficiency for your independent agency through the intelligent use of technology.


Share your thoughts in the comment section below, subscribe to get updates delivered to you and *please share this if you found it informative.

Transcript

Michael Jans: Hey Chris, how are you?

Chris Cheatham: I’m good. How are you doing?

Michael: Well, I am good and I’m excited about this conversation. Before we dig into the juicy stuff, you’ve got an interesting background and career so tell us a little bit about how you got to be where you are and then we’re going to take a look at how you’re solving problems for independent insurance agents.

Chris: Yes, happy to do it. I used to be a surety claim lawyer. I like to say that if you are an agent, you did not want me showing up at your customer’s door because it means something’s gone very bad.

Michael: [laughs] All right. Where did you do that?

Chris: I was in Washington D.C. when I was doing that.

Michael: All right. I interrupted you. You were a surety lawyer and then what happened?

Chris: There was a case I worked on for many years and the documents were a complete mess when I was going through them for litigation. I just remembered thinking at the time like, “Man, there has to be a better way to do this.” We initially launched our company in 2012 to basically organize, collect, and deliver claim documents in software, so we were working with attorneys, accountants, and consultants to collect and access claim documents. Then in 2014, we had an underwriter login because it was a very big claim so underwriter logged into our software, she called me and she said, “Hey Chris, can we use your software for a policy review?” I said, “Sure. What’s policy review?”

Michael: [laughs] Okay.

Chris: I always laugh at that because I had no idea what I was getting myself into. It’s kind of the rabbit hole that created RiskGenius.

Michael: Boom, you’re in Wonderland now.

Chris: Yes.

Michael: Okay. That was your insight or the inspiration to start the company that you started.

Chris: Yes. That was one of the many insights. After I got that call, I started calling some agents and said, “What’s going on with this policy checking thing?”

Michael: What’s this policy review thing?

Chris: I’ll never forget I have a friend who worked at a big brokerage. She said, “Oh, here’s my weekend”. This is what I did last weekend. I sat in a conference room with another person. We both had two versions of the same document. I read one version to her out loud and she red lined the second version.

Michael: Oh, my God.

Chris: I was just dumbfounded. Then there’s another person I talked to. This person was very technologically-advanced so I said, “How do you review policy language right now?” She said, “Well, I print out the documents that I need to compare, I pull out an exacto knife, and then I cut the policy up into clauses and then I moved the clauses around on my desk.” [laughs] I just was like, “Okay.” We’ve found a problem to solve.”

Michael: Okay, all right. Very briefly, how would you define that problem?

Chris: The problem is nobody has any clue what’s in their insurance policies. I don’t mean that flippantly, like a broker that is reviewing an insurance policy or has placed an insurance policy certainly is going to know what’s in that policy. But like the macro level, people really just don’t know what their coverages are, what their exclusions are, they don’t know what they should be looking for, they can’t keep up with the demands of doing policy-checking so we wanted to fix that problem.

Michael: Got it, all right. In a moment, I want to talk to you about how you’re solving the problem. Before we do that, I want to ask you from your perspective– You’ve been running an insurer tech essentially for, if I’m doing the math right here, about four, five years?

Chris: Yes, that’s right.

Michael: Okay. To some extent, you’ve been an observer of the industry longer than that?

Chris: Yes.

Michael: Right? I don’t know how long you practiced on the surety side of the game. You’re not new to the industry but you may not have been in the industry as long as I have. What do you see happening right now? Why is it that, for example, a company like yours can appear whereas, let’s face it, if it were 2005, you wouldn’t have started that company. Right?

Chris: Exactly, really simple. Yes. For our company specifically, the reason why we can exist now and we couldn’t in 2005, there’s two reasons. Number one, is the cloud. This is the same for every industry that’s seen itself going through a tech revolution or evolution or whatever you want to say. With the cloud, a lawyer like me can go hire some developers and build software and put it up and make it available, secure, and accessible. Right?

Michael: Yes.

Chris: Before, I would have had to go and buy servers and figure out how to spin those servers up. It was just a big mess. The cloud changed that side of the equation. The other thing that has changed is the insurance industry was one of the last industries, in my opinion, to go digital. By digital, I mean just simply going from paper documents to electronic documents.

When we launched our company back in 20… what was it, ’14? When we launched RiskGenius I had a friend who worked at a big carrier, a big international carrier and part of his job every Friday for a couple hours, he would sit at a scanner and scan documents. He had an MBA, he is a professional person doing scanning because they were trying to take all their filing cabinets to make them digital. That’s what they came up with.

Michael: That was breakthrough stuff back in 1995.

Chris: I know but a lot of these carriers and there are brokers still out there that are not digital at all, everything is still paper. Until you get everything digital, you couldn’t really do things that were  [inaudible 00:09:29].

Michael: You’ve observed other industries and you’ve worked in other industries, why do you think the insurance industry has been so slow to adopt digital technologies?

Chris: I equate the insurance industry a lot with the banking industry, it’s just that insurance–

Michael: But banking really went Fintech before insurance went in InsureTech, right?

Chris: Totally, absolutely right. I’m not sure why banking went before insurance, I’m sure someone much smarter than me would understand that-

Michael: Fair enough.

Chris: -but I think those two industries lagged other industries because they were just more about moving capital around than they were about having extremely efficient processes. When the key driver of a bank is the capital sitting on their balance sheet, which in all intents and purposes, I would argue, that’s the key driver of insurance companies too, is making sure they have enough in reserve to cover what they’re taking on.

When that’s the key thing you have to think about, you focus on the financial side, the monetary side, and then your IT systems are a little bit secondary. As a result, you didn’t have to worry about quick processes, you had to worry about me being efficient with your money. That’s where you spend money on actuaries and that sort of technology as opposed to processes and workflow technology, for instance.

Michael: Okay. When you look at this industry changing– because when I ran an insuretech myself, while it did seem to happen fairly slowly in the beginning years, in retrospect, I can look back and say, I saw the industry change, I saw increasing acceptance among my marketplace, which was the retail independent insurance agency. I saw increasing acceptance of digital technologies, it was different when I started it 10 years ago than perhaps five years later. Are you seeing an increased interest in acceptance in various digital technologies?

Chris: Yes. 100%, and actually, I always go back to one of the very first pitches that I really ever did of our company. I was talking to an executive of a big carrier and he actually had one of his outside counsel was present for the meeting too, and so I am doing my pitch and at the end, they say, where’s this hosted? Where do you put the data? I was like, “We use AWS to host our data into our web services portals. The executive say, “We’ll never do that, we’ll never go to cloud.”

Michael: Okay. I remember those conversations. I remember early conversations about that. Okay. Keep going.

Chris: It got better because the general counsel who’s sitting next to him says, “Yes, we don’t want to use the cloud ever. We don’t use the cloud either.”

Michael: We don’t trust the cloud. We trust the server in my closet.

Chris: Actually, he said, “We don’t trust the cloud but you should check out Dropbox.”

Michael: Okay, right.

Chris: I almost started laughing out loud because what he didn’t realize his Dropbox is the quintessential cloud application.

Michael: Obviously, in the cloud. Okay, “Check out Dropbox.”

Chris: Yes. “Check our Dropbox.” That was pretty funny. It is totally changed, now it’s about what server are you using? AWS passes every test. Some of our bigger customers, and this is more on the carrier side focus on having a private instance of that cloud solution. If you’re an AWS, somebody might say, “I’d like this to be in my private instance of AWS,” which is workable, but I don’t hear the cloud as an issue. Almost never. There’s no one customer and I’m thinking of where it’s still an issue, but otherwise, it’s not a big deal.

Michael: Got it. Okay. Before we dig in to the weeds or the way you’re solving problems, I want to ask one other bigger question or bigger picture question. Now, when you look at the independent insurance agency, let’s say, of the future, whatever the mysterious year is, three years from now, five years from now, how do you perceive it being different, functioning different, serving it’s customers differently than it is now? What does the agency of the future look like to you?

Chris: I’m going to focus on one specific area of the agency of the future, I’m going to focus on insurance policies because that’s really what [crosstalk]. I feel like I think what people need to realize is that the brokers are in the driver’s seat to dictate the terms of insurance policies. What I mean by that is like, one insurance carrier has that carriers insurance policies. They know the language that’s used, they know that pricing they use, but they don’t know really what their competitors are doing. Whereas a broker has all of that data, every single bit that they need because they have policies and quotes and binders from every carrier that they possibly can talk to.

The reason why that hasn’t been super powerful in the past is because the brokers can’t structure the data that’s in those insurance policies in an efficient manner, it’s really hard to go in and type out all the limits and type out all the forms and do all that work that’s necessary.

With artificial intelligence, with machine learning, that data is going to become structured very, very efficiently and as a result, a broker of the future is going to be able to sit down and say, I have a construction firm of $50 million in Kansas City. Here’s the last policy that I placed from a construction firm of a similar size in the Midwest and here are the endorsements that we always want to get, and here’s the underwriter I should go talk to, and here’s the limits and look at this, we just got this limit down from this particular underwriter, so I should go to him for this particular thing. Whereas right now, you’re just going to the same person for everything. That’s the future in my mind.

Michael: Okay. I’m going to circle back on some technology related questions, among them– I’ll just put this in the back of your mind because as we get more digital as an industry, it seems that the agency principal needs to make intelligent decisions about which technologies to invest in, which ones not to invest in. They need a buying process for something they likely haven’t bought before. In the back of your mind, one of my current projects is identifying a good buying process for technology. We’ll circle back to that one. Let’s talk about RiskGenius. You started this company, five-ish years ago. I want our listeners to know that this is a company with substance. You’re in– is it Series B funding or–?

Chris: Yes. Series B funding.

Michael: This is a company that’s well recognized by the, in this case, Venture Capital Community. I think you’ve got carriers as investors as well, and clearly you’ve got clients, which is, of course, where the rubber meets the road. You had an insight that the policy review process was inefficient, cumbersome, probably not up-to-par. You chose to jump in and help solve that problem. Tell us a little bit more about, how does it work? What does it do? What does it mean to the independent insurance agency of today?

Chris: I’ll start how it works. We’re big believers– we’ve done a lot of user testing, a lot of user feedback sessions. One thing we realized when we started trying to tackle policy checking is that we didn’t know anything about policy checking and we really didn’t know anything about insurance policies. We learned that the hard way, our first launch back in 2015 because we had built algorithms for commercial general liability policies without really fully realizing that there’s other lines of business, which I know is extremely foolish, but at that time, I did not know that was extremely foolish.

Michael: I’ll jump in with a quote from Reed Hoffman, founder of LinkedIn. “If you’re not embarrassed by your first launch, you waited too long”.

Chris: Yes. That’s exactly right. We people jump into the software when we launched and they started uploading property policies. I was like, “No big deal.” But then we got to the sections that included perils-insured and perils not insured and business interruption, and it was stuff I’d literally never heard that before–

Michael: And a bunch of other insurance words that the normal citizens don’t understand.

Chris: Yes, and we quickly realized that not everyone uses ISO policies, it was just a nightmare. We’ve had to do just so much work around learning to understand insurance policies, then learning how people read insurance policies and analyze them and then finally training– taking data scientists and showing them that step of how to read an insurance policy, and then writing the algorithms that show how people read insurance policies and then you have to build on the back-end, which we lead, jump-start on because we’d already built a document management system.

Then you have to build in the user interface that goes on top of that. It has been a long, hard path, I’m not going to lie, of figuring this stuff out, which it should be because, honestly, if policy checking was easier, somebody would have done this a long time ago.

Michael: But you’ve got five years into that now?

Chris: Yes. Absolutely. Yes, we’re just at the point now where we launched in October a solution that does not require what I call human intervention. That’s really the question you should always ask about any artificial intelligence solution. How much human intervention is occurring right now? I think about–

Michael: You mean, when I speak into my Google Home, I know that engineers are secretly listening to my questions?

Chris: That’s exactly right. There’s an algorithm that sorts out all those words that you say and they try to understand them. I actually know somebody who was working on an open source version of Google Home. What he was doing with the words that people are speaking into devices is no different than what we’re doing with the words that are appearing on an insurance policy. You’re trying to parse and understand them and be able to teach a machine how to parse and understand them. That’s what we spend a lot of our time on.

Michael: You hit a tipping point in October where now you’ve got– Just explain that to us.

Chris: The first version of our software, which came out in 2015, required humans to go in and help the machine figure out what it was looking at. Either we were helping identify, “Oh, here’s a clause”, or we were correcting labels that were getting attached to clauses. Imagine there’s a terrorism clause and the machine accidentally identifies it as a war clause. We have to go fix that. Or imagine a terrorism exclusion, and a war exclusion get combined together, those had to be separated out by a human.

The goal is to reduce that type of human interaction. We hit this tipping point in October where we don’t have to have that human intervention anymore because we’ve done so much work on the front end to make sure the machines understand what they’re looking at.

Michael: This is fascinating, and I just want to ask you, if you can, what did Denzel Washington say in the movie, Philadelphia, “Talk to me like I’m 10.” Artificial intelligence, tell us a little bit about how AI works and how it learns.

Chris: That’s a very broad question.

Michael: Talk to me like I’m 10.

Chris: Talk to you like you’re 10?

Michael: Well, because clearly your AI is a little bit smarter now than it was in 2015.

Chris: Yes, absolutely.

Michael: How did it get there?

Chris: Let me imagine you’re a 10-year-old, and I’m teaching a 10-year-old how to do policy review. I start out with one policy and I say, “Okay, Jimmy. Here are five things you need to check for, you need to look at the named insured and the limits and the deductible. Let’s just do those three things.” I go through and I show Jimmy exactly what they are. Highlight them, then I put a post-it flag on it and I say, “Named insured.” Then I highlight the next one and I say, “This is the deductible.” Then I highlight the last one, I say, “This is the limit.” Jimmy’s like, “Okay, great.” There’s one example.

Then we go to another carrier and I say, “Now Jimmy, show the named insured, the limit and the deductible.” He gets the named insured right, he misses the limit because the table got called something else and he misses deductible because it got called something else. I help him identify those. I correct what he’s just done and so he’s learned from that. Imagine that process over and over and over again until Jimmy’s seen enough that he actually can take a new policy he’s never seen before and find the information.

Imagine I give him a final test, and I say, “Okay, Jimmy, here’s 100 policies. You now have one hour to go through these and check for those three things.” Jimmy comes back and he’s 93% accurate on the name insured. He’s 95% on the limit and he’s 80% on the deductible. I say, “Okay, Jimmy, great job. You’ve passed the test for name insured, you’ve passed the test for limits. We need to do more training on deductible, so, I’ll find some more examples and show you those.”

That’s literally it in my mind. I know it’s a lot more complicated, but that is the process. Then there’s ways that you take that information, and you teach machines to understand it. That’s where the algorithms and the different ways to do artificial intelligence come into play. Frankly, that is not my expertise. I’m not going to even make up stuff.

Michael: Fascinating, nonetheless. Talk to us about the problems you solve for the independent insurance agent or broker.

Chris: The key problem that we initially tackled was policy checking that occurs year over year, particularly around renewals. Taking the last year’s policy and comparing it to a quote or taking the policy that gets issued and making sure the quote matches what came out in the final policy. Or doing something along those lines or adding a submission in there. Policy checking, it should happen on every single account, everything single year. It doesn’t always happen that way inside brokerages and agencies, but it is what it is.

Our software helps people do that much faster. You have 100-page document, you could go through all 100 pages and do the policy check manually or you can use our software and you click on each item or split the check, it takes you to that section of the policy. You do your quick review and then you’re on your way. It takes two minutes and 45 seconds on the demo that’s on our website to do a policy check that would take an hour typically in the past.

Michael: I was going to ask you, do you know how much time you save somebody?

Chris: Yes, a lot. Brokers out there probably know that the policy check you do depends on how big the customer is and how big that policy is. Sometimes you get more–

Michael: Maybe how diligent you are with your policy check?

Chris: 100%. You might have a very manuscript policy, and you actually have to do more analysis of that to make sure it’s correct and you’re getting what you expect, and things haven’t changed year over year. Whereas a mob policy or simple package policy, those can be done pretty quickly with our software. The manuscript can be done very quickly with our software, but you’re talking anywhere from two to four minutes on a basic policy. I think the bigger more manuscript policies where you’re checking the clauses and the policy language, that’s going to take a little bit more time because you got to go and read the clauses.

Michael: Got it. Alright. Is it commercial lines only?

Chris: Yes, commercial lines only. Many people just don’t really care what’s in the personal lines–

Michael: It’s commercial lines of all sizes or–? I think you mentioned BOPS.

Chris: Yes, all sizes.

Michael: All sizes, not just middle market. Alright. Typically, you see your solution being used around renewal, right?

Chris: Yes.

Michael: A broker may have secured a quote from perhaps two or three other companies than the incumbent company. They want to be able to provide a comprehensive review for the customer.

Chris: That’s exactly right. Yes.

Michael: Now, let me ask the next question. Can they also effectively do this in a competitive situation so now Mr. Broker down the street is also competing for your business?

Chris: Yes, absolutely.

Michael: As long as I get a copy of that policy or how does that work?

Chris: Yes, if you get a copy of the policy, that’s great. Oftentimes, we can also work just with a deck page, if that’s all you’re getting it for, even just a quote, if that’s what you’re getting. What we do, let’s say, there’s a producer that’s really good at going and talking to manufacturing companies, and they like to do coverage gap analysis. We will sit down with that broker and we’ll say, “Okay, what do you check for?” It’s really the IP of that particular broker. What’s your intellectual property?

We then take that information and make it into a checklist that is customized just for that producer. They can go in, have their customized checklist, upload the, let’s say, the deck page or a quote or policy from the prospect. They upload that document, check for the things that they need based on the checklist they have on the left, and they’ve done their review, they can export out the results and share with the customer. It really speeds up that process. Frankly, that’s a process that gets ignored sometimes because you have to do policy checking while the account managers are busy. We think we’re empowering people to do that a lot more.

Michael: All right. Chris, let me do a reality check on the solution. It seems to me that there are two or three significant benefits from the solution. One, if I were a broker, I could deliver a comprehensive policy review with confidence to my client. In other words, I could really demonstrate to them that I’ve done a comparative review. Right?

Chris: Yes.

Michael: Two, I think it probably imparts a greater degree of confidence to the customer to the end user. Three, in some significant likelihood, I may be able to identify gaps so that I can sell more appropriate coverage.

Chris: That’s all right. Yes. There’s a fourth one, too.

Michael: I know, I’m not done yet. I got one more and then maybe you have one. Four is that I can steal customers from my competitor. Did you have another one besides that?

Chris: I did. I have one more.

Michael: There’s five? All right.

Chris: There’s five. There’s probably more than five, but we’ll discuss five today. The fifth one is what I call policy optimization. You’re doing your day-to-day thing, where you’re checking the policies or checking competitor policies et cetera. Over time, because you’re doing a standardized checking process, we are accumulating data for you about what is good and bad about your policies and what is good and bad as you’re doing your review, and so that data can be used in a number of ways.

One way that data can be used is you could go and say, “I want to look at all the red flags. Everything that I flagged on travelers renewal policies. I want to see what they’re screwing up every time.” Let’s say they’re messing up the named insured every time. You can turn around to that carrier and number one, you could just say, “Fix this.” Number two, you can say, “I’m having to do inordinate amounts of work to correct your policies. I want a better rate.”

There’s all the different things you can do with that data or there’s also concern off-siteoutside of your politics. Construction customer example, you can start looking at construction policies and seeing 98% of the time, we’re going and getting our named insured endorsement. What about these 2%? Why aren’t we doing it here? Are those mistakes that we’re making? Is that going to create E and O? We have to fix that going forward. That’s the other piece.

Michael: It really does support efficiency. Let me ask you one last question on this. At your level at the insuretech, the RiskGenius level, I would think that your AI is learning things industry-wide not just agency-wide. I don’t know if you’ve been able to do that yet, but at some point, it would seem that that could turn into extremely valuable information for the industry.

Chris: Yes, absolutely. We’ve done this in the past actually where we survey the policies to understand the content. I think that small brokers will band together for example and start bench-marking their policies against each other, but absolutely. This data can be anonymized, and we don’t do it without people’s permission. I think this is very, very important. We’re not going to take your data and sell it to somebody else, but there’s some very interesting things that can really help the insurance industry going forward as long as people are working.

We’re willing to work together with this data. There’s also trade associations– there are groups of brokers that are out there and have an interesting seat to their member brokers working together. Really interesting topic that’s going to start out of all this data. I’d say that’s the second wave of insuretechs that you’re going to see or insuretechs that are doing really interesting stuff with data, I’d say the previous insuretechs you’ve seen, I just thought, NDO, for example, got bought by Applied. They’re fixing the piping is the way I think about it.

They’re making the submission process better. There’s just a lot of companies trying to fix the pipes, and then there’s companies like RiskGenius that are now trying to send different data through those pipes so it’s more easily understandable.

Michael: Fascinating. Now I want to circle back to the question I put in the back of your mind. On a daily basis, you’re dealing with perspective customers who need to make a decision about purchasing something they never purchased before. While every great company– you’ve got a sales process just like the insuertech I ran had a sales process. One of things that we quickly encountered was that our prospective customers did not have a buying process for technology. What counsel or advice would you give to independent insurance agents about how to make decisions about purchasing technology?

Chris: I would say do not make decisions from the top-down. You have to get your users involved, otherwise you’re going to buy stuff that you don’t need. I was actually just on a call this morning with someone, and it was a manager of a bunch of policy checkers and she was skeptical of our software. She had users on the call this time, and there was this pause at the end where she’s like, “What do you think?”

The user just goes, “I love it. This is exactly what we need.” I have seen this gap too many times where the decision-makers aren’t necessarily the users, so they don’t understand the problem that needs to be solved. As I would say, the key to buying a stuff, in my opinion, is engaging your users to understand where the pain points are. Don’t buy something just because it’s the hot, cool insuretech out there. Don’t buy something because it’s the buzzy technology word out there. Buy something because it fixes a problem. Be sure it’s going to fix a problem until you engage the users and let them try the software. That’s really, really important.

Michael: Chris, I have one last question for you. I want to go with the insurance highway question because I think you’ve got an interesting perspective, so if you had the opportunity to put a billboard on insurance highway and everybody’s driving let’s say 75, 80 miles an hour, and you could put a very, very short message on it that it’s non-commercial to RiskGenius, but a message you want to deliver to the principles of independent insurance agents today. What do you want to say? What do you want them to wake up to?

Chris: I think the message should be something like “Own the Data”. What I mean by that is your data is very, very important, and you need to understand what you have and what you don’t have because that’s going to be the key in the future. I would just say focus on the data because that’s where everything’s going to be in the future.

Michael: That data as its representative presumably of their customers.

Chris: Yes, absolutely.

Michael: Got it. Chris, if our listeners want to find out more about RiskGenius or ask follow up questions, how should they do that?

Chris: You can go check out our website, riskgenius.com or you can just email me chris@riskgenius.com. Either way works. I also post on LinkedIn a lot, so if you want to come see me harass people about what insuretech is doing good and bad, it’s a good place to do it.

Michael: Very good. It was an exciting conversation. I appreciate you sharing your insight with us.

Chris: Absolutely. Thanks for having me.

Michael: You bet.