Sound Policy podcast: AI transforms insurance claims

Welcome to Sound Policy from FM. On this show, engineers, researchers and insurance professionals break down the challenges businesses face and the strategies to overcome them. Today we’re talking about one of the most important parts of insurance: claims. New technology, like AI and automation, is changing claims in significant ways. To learn more about what the future holds, we sat down with Jeremy Gallant, senior vice president of claims at FM. 

Brian Amaral: Jeremy, thanks for being on the show.  

Jeremy Gallant: It’s great to be here.  

Brian Amaral: First, introduce yourself. Tell me a little bit about yourself. What do you do at FM?  

Jeremy Gallant: 20 years at FM. A couple of years at a different insurance company previous to that. New into claims in the last couple of years. Came from an underwriting background. So, I'm the SVP of claims, managing our global claims portfolios.  

Brian Amaral: What excites you about it? What motivates you to get up every day and do that work? 

Jeremy Gallant: Well, that's a great question. Every day is a different day. You never know what types of claims are going to come in. You get to see all of the businesses that we write pretty much around the world, from big losses to small losses. You get to work with people all over the world: colleagues, our brokers, our clients. So, you just see how the world works. 

Brian Amaral: I imagine there's a lot of times when you're working with people going through a really difficult professional experience of dealing with especially large losses. How do you handle that? 

Jeremy Gallant: That's a great question. I mean, I think at the end of the day, this is a people business. So, we try to do as much work pre-loss, as we can. So we like to know our clients, make sure we know who we’re going to be negotiating with, make sure they understand what our policy covers and what it does not cover and, you know, really work with underwriting in advance to make sure we're tailoring unique coverages that they need so they know they have that certainty of what to expect if there is a big loss.  

Brian Amaral: So 20 years at FM. What would the senior vice president of claims, what would their desk have looked like 20, 25 years ago? 

Jeremy Gallant: A lot of accordion files, papers stacked up. And pretty much now I can work from anywhere, anywhere on the planet, on my laptop or tablet. And everything's been condensed into, you know, email or other devices be able to do our jobs.  

Brian Amaral: What has that meant for FM’s claims service, that move to digital? What has that meant?  

Jeremy Gallant: The huge investment that's happened, but I think that: A) It gives us an ability to collect our data in a much better way that's not on paper, so we can use it in various fashions. Helps us get to our clients in a much more expedited manner and helps our adjusters to focus on the things that are important, which are helping our clients get back into business following a loss, and hopefully doing a lot less admin type of work than they would have done previously. 

Brian Amaral: Has that change been accelerating over the past few years, especially around AI, all this additional technological transformation? 

Jeremy Gallant: Yeah. I think claims has been one of the benefactors of AI and the changes that are taking place. What wasn't possible last year is possible this year. We've looked at our entire claims process from our intake and our first notices to our payments. And just looking at what administrative steps are in here and how do we remove that from our adjusters’ hands so they can spend more time with clients, either pre-loss or out on sites, helping our clients recover as quickly as possible to get their business back in order and meet their clients’ obligations. So, I see that continuing. Computers and data, pretty much every step of the claims process has those commonalities in it. And whether it’s automation of process, damage assessment reviews, triaging our losses, how do we handle big catastrophe-type claims, how do we measure adjuster productivity—what should they be doing—and then how can we automate as much of the process as possible, realizing that we still believe this is a people business, and we really need to make sure that it’s still human to human working. But how do we remove some of the administration tasks from the adjusters to be able to concentrate more on getting with clients and getting claims settled quickly and fairly.  

Brian Amaral: How do clients benefit from that approach? 

Jeremy Gallant: In some cases, on small losses, it could be less requirement for information to get paid. We've introduced a new process called fast-track claims to expedite small claims. So, we kind of did a full review of, okay, what critical pieces of information do we really need to adjust this file? And then we’ve tried to reduce that to as much as possible and then send those out on an automated system. So, once clients receive those, they upload the documents, and a human is still looking at them to make sure there's no fraud and the estimates that they receive are accurate. And then the claims are paid. So, in our clients’ sense, this is in their control to how fast this gets paid. We started this last year; we’re now reduced cycle time by up to 50% from what it was previously.  

Brian Amaral: Cycle time: what do you mean by that?  

Jeremy Gallant: Cycle time is, you know, from we're notified of a claim, how long does it take to get that claim settled and paid. 

Brian Amaral: So a client who has a loss, a lot of them are getting paid more quickly as a result of this system?  

Jeremy Gallant: Right. So, say was 180 days on average before, you know; it's now 90. And, you know, there's no reason that can't be 60, for small losses. 

Brian Amaral: Take me back to the decision to launch fast-track claims. Why did you decide to do that? 

Jeremy Gallant: Yeah. Great question. So, when we were looking at all of our loss experience, we noticed, you know, about 80% of our losses come from amount that make up a small proportion of our overall claims payments. So it's a lot of the volume comes in, don't really affect the outcome of the company. So, we wanted to take a step back and say, “Okay, we’re hearing from clients. They want to get paid faster. There’s too much administration in adjusting some of these small losses.” If you think of losses under $100,000, under $50,000. For the most part, clients go out, they get estimates. We don’t quiver over those amounts to a large degree, so why don’t we just pay them? If they supply and they did it, and here’s the estimates, then we just make that process a lot easier than it used to be.  

Brian Amaral: In the reverse of that, the higher dollar figures, those are still handled with a white glove service that's not in the fast-track system. How does the fast-track system benefit those sorts of losses as well? 

Jeremy Gallant: So, what's happening is that we're saving in the 300% of adjuster hours by automating most of the fast-track stuff. So, what that means is our adjusters have more time to be able to spend with our clients, whether it be pre-loss, better understanding the policy, working with them on tailored solutions, policy workshops, and also more importantly, you know, how do we add value when there is a big loss? How do we get on site as quickly as possible? How do we be there with the clients as they’re starting to make decisions right away following a loss. And we’re finding that our involvement in early stages within that is helping mitigate these losses very substantially, which means our clients are getting back into business faster, which they’re able to meet their clients’ expectations. And making sure we can do that with speed and clarity and the expertise that we bring to the table has been phenomenal and I think only going to get better.  

Brian Amaral: How does faster claims impact the client on a day-to-day basis? What benefit do they see from that system? 

Jeremy Gallant: So, it's totally within their control to get those claims back submitted as fast as possible. And they’re in control of how fast they get paid. So, as long—we set out from day one what requirements, what documents, what reports we need—and this is a list of four or five items; it’s not a lengthy list. So, once they upload those, pretty much almost instantaneously they can be paid and the file can be closed.  

Brian Amaral: Claims seems like a data rich environment. It's a people business, but you also have a lot of data. Can AI help bring that data out and inform other areas of the business as well? What can you learn from that claims data with AI that you maybe couldn't have back in the days of the accordion file?

Jeremy Gallant: For us as a company, I mean, we're not a high-volume claims thing. So, you know, large losses, it's hard to, you know, what happened to this industry may not be the same that’s going to happen in  other ones. But our core losses, still as I mentioned earlier, was 80%, it’s small losses, which we define as under a quarter million dollars, which is still fairly large. There is a lot of data that can be extracted there, and can be used for underwriting and engineering in order to make better loss predictions upfront. It gives us a thing, where's inflation going? Where is supply chain going? Are we seeing longer downtimes? Et cetera. 

Brian Amaral: To go a little bit more general here. When you're, when you're talking to people about your work—you're at a party, you say, I'm Jeremy Gallant, senior vice president of claims. What's something that people just commonly misunderstand about your work?

Jeremy Gallant: I think anyone says if you say you work in insurance, it's little skeptical right away because everyone thinks of their auto insurance or their home insurance, that maybe have had a problem over the years. And, you know, we insure the best, biggest companies in the world, helping them manage their business, helping them be more resilient, is our key focus. While we do believe the majority of losses are preventable, it doesn't mean all losses are preventable. Natural catastrophes continue to increase in size and scope. Big fire losses, big equipment losses, certainly those are continuing to see upticks. And large exposures around the world, and supply chains are shifting due to geopolitical reasons, war issues, conflicts, lots of different reasons. And us being able to respond to those is very interesting. So, we get to see how the world works, how companies work, how supply chains work, how businesses work, how clients make money, and how we can help them get back into business should something happen. 

Brian Amaral: That sounds more interesting at a party than they might expect. 

Jeremy Gallant: People are definitely surprised. 

Brian Amaral: Do you think there could be a time where AI would replace human decision making? Or are humans always going to be in the loop in this claims process?

Jeremy Gallant: I think humans will always be in the decision-making process. Now, does that mean every decision that’s made today is going to be made by humans or…I don’t know. I think that this will allow claims to get more impactful decisions to the right people at the right time. Decisions that everyone’s going to agree with or decisions that are fairly simple, let’s not dwell over those. Those shouldn’t hold up the process. Let’s have those made, and let’s make sure our people are handling, you know, the value-add, important, strategic decisions that need to be made on a claim.

Brian Amaral: What are some of the barriers to AI really helping streamline processes, to really improving things?

Jeremy Gallant: I think data quality is one big one. So, I mean, if you don’t trust the data, then obviously, decisions of AI using it isn’t going to work very well. And the models that they use need to be proven and tracked and predictable. And, you know there is regulation out there that exists to make sure we are not relying on machines to make decisions that impact people’s livelihoods and businesses. So, I can see more coming there, and that’s why we’re not trying to remove the human from the process. We’re just trying to direct the human to where they can add the most value. 

Brian Amaral: What are clients telling you about this? Especially fast-track claims, which is now out there in the wild. What’s the feedback from clients been?

Jeremy Gallant: It’s been amazing. I mean, I think at the beginning, there’s always a new process, and people don’t understand it. So, I think at the beginning, we did, said, “Hey, you know, we used to have an adjuster come out, and they would help us through this.” But then when you talked to the clients, it’s like, “Okay, well, what did they help you do?” “They helped us make the decisions.” “Okay, and what decisions did they help you make? And would they have been different than you would have otherwise made?”  And so we started working through those, and it’s like, “Oh, no, I think you’re right. We knew how to do this. We just wanted to make sure we were going to get paid.” So, a lot of what this is about is to make sure the trust is there and that the process is going to work. So when, you know, clients do follow what we’re asking them to do, we are going to pay at the end of the day. And that’s, I think, a lot of trust that’s been built into the process to make sure it’s worked. So, we did this in the US and Canada about 18 months ago. So, I think we went through that, and it’s really stabled out quite nicely now, and we see positive feedback pretty much from everybody now. This is taking a lot of work off brokers. That I used to have to follow up on these $50,000, $25,000 losses; that’s not the best use of my time either. So, if clients are happy and they’re getting paid what they believe fairly and they’re getting the money faster, then, hey, this is a win-win for, for everybody. We recently launched the same automated method that we did in the US in France and Germany, where we do see significant volumes of small losses. And, you know, 95%, you know, we’re getting there on the same feedback. I think as any new process, you know, it just does take a while to get everybody up to speed on what we’re doing, why we’re doing it, and how we’re helping them get their claims settled fairly and quickly.

Brian Amaral: Are there any specific examples that you could point to, any anecdotes that really stand out to you as part of the fast-track claims process?

Jeremy Gallant: Yeah, we have a European client. I mean, they have, you know, 300 losses a year. I mean, they're mostly all small. And you look at the amount of time that was spent between everybody touching that, whether it be their client, the local facility that had the loss, the corporate client, the broker, us, our claims coordinators. I think having that mostly automated at their control has dramatically sped up their claims portfolio and is getting them a lot more claims quicker. And we’ve had other clients where in the past, we we've kind of shied away from writing the business because they knew that there was a high volume of low claims and we just didn't have the resources to be able to effectively manage it the way that we did. So now, you know, it's leading to some new opportunities out there that a couple years ago we wouldn’t have been able to do.

Brian Amaral: Oh, interesting. What excites you most about where claims technology is headed in the next couple of years?

Jeremy Gallant: I think it's bringing back the value that claims can bring to an account. You know, you think of an insurance company, and at the end of the day, we sell a policy that's going to pay for things that happened if it’s covered after an event. And we want it to be as broad as possible and interpret it as broad as we can to help our clients get back into business as quickly as they can.

Brian Amaral: Is there anything that’s really surprised you about AI and automation in particular since these tools have started getting either piloted or rolled out? Anything that really stood out to you as a surprise?

Jeremy Gallant: I just think about how fast some of this has really happened. You know, what were concepts two years ago are implemented now and working. And, you know, I think a lot of people try different AI things and, you know, maybe it works, maybe it doesn't work and how fast can we pivot? I think we took maybe a little bit longer to get things going, but a more planned thing to make sure that, hey, once it was released, it was going to work. Because if you change your entire process on how you handle 80% of your losses, you’d better make sure that it’s going to work well. Which I’m proud to that say it is working well. And there’s things that we’re learning, and I think it's something that's always going to be evolving. And we are always taking our clients’ and brokers’ feedback, and trying to make the process more streamlined, better and more predictable, and making sure that we’re responding to the needs that our clients and brokers have. 

Brian Amaral: What's one big takeaway that you want listeners to really remember about AI and automation and technological development, how they're going to impact claims?

Jeremy Gallant: Yeah, I think, you know, it's taking away the administration side of the job. It's not taking away someone being there for you during your loss. So, we are still doing that, and we're trying to do it a lot better than we did before. And you know, we do want to make sure this is enhancing the claims experience, not detracting from it, and we’re providing more value along the way, to say why do you buy insurance from us? Because we are going to help you get back into business faster than anybody else will in the industry.

Brian Amaral: Claims is a people centered business and is going to remain so, even if that person is using AI to improve their job. 

Jeremy Gallant: 100%. 

Brian Amaral: We started this conversation talking about what your desk would have looked like 20, 25 years ago. What about in 10 or 20 years? What is, what is the senior vice president of claims, what is their desk going to look like with this AI future that's upon us?

Jeremy Gallant: It's hard to tell. I mean, I think, you know, again, where we were last year is not where we are this year. And next year, who knows what's going to happen. So I think it’s only exciting. You know, you could see data being able to be supplied to, to be able to get people to make strategic decisions. So, where some of our adjusters on medium to large losses could handle 40 losses today, maybe they would be able to handle 100 losses. Not that they will work more than they work today, but they will just be able to be funneled and be able to work on different areas of things where they add the most value. 

Brian Amaral: What can go wrong if somebody is using AI in the claims adjustment process?

Jeremy Gallant: Yeah. I mean, if you don't have rules and procedures around how you use it, AI can take you down a path that may not be accurate. So, not knowing how to prompt things properly or researching the wrong information. It could draw on things on the web that that may not be accurate, or may have been accurate ten years ago that aren’t accurate today. So, I think there is a lot of knowledge that goes along with this to be able to look at it and have enough education of the process to know what is correct and isn’t correct, or at least enough inkling or ability to be able to question something if it seems off. And I think that is why some of the regulation around this stuff is not crystal clear yet of how far insurance companies should go with AI. And we view this as important that our adjusters are trained in this, that they can they can adjust the files. Yes, AI is going to supplement and allow them to make decisions. But at the end of the process, they are still the ones making the decisions. 

Brian Amaral: Yeah. If you ask AI for dinner recommendations and it sends you to a closed restaurant, that’s one thing, but if it tells that you a claim should be accepted or rejected or adjusted in a certain way, that probably has more implications for our clients. 

Jeremy Gallant: Correct. 

Brian Amaral: What have the early results shown about these initiatives? 

Jeremy Gallant: Yeah, I mean, we've looked at our, you know, from an employee standpoint and from a client standpoint. So, from our employees, we've done surveys of job satisfaction and we’ve seen some improvements there. We’ve looked at, you know, how much time they spent on various tasks before and after, and it all shows positive improvement. But I think more importantly, on the employee side, we've kind of made them part of our AI strategy. So, we’ve had a group within our claims department; they came up with some of the ideas that we’ve looked through, say, okay, can we implement these things. So, we're getting a nice bottom up, I guess, if you will, approach to ideas, where our adjusters are spending their time, where they don't feel that that's the best use of their time. So, we’re not solving every single one of them for, for different reasons. But we are getting out there and solving the problems that they are telling us are issues. On our client side, I think the same thing. You know, we have changed how we’ve handled our smaller losses. So, I mean, we are looking, you know, surveys based on the experience. The same with brokers, to figure out what has changed, how’s the process gone?  Is there feedback that they would like to provide? And most of that has been going very well. And as you remember, we talked about earlier was that our cycle time is half of what it was and there’s no reason why it can’t continue to improve. So I think that is having massive benefit for many of our clients who just want to get losses settled and paid as quickly as possible. Gets off their desk, they’re happy with the outcome, and they get back to doing the business that they want to do. 

Brian Amaral: What are some of those administrative tasks that AI and automation are already helping clear out of the way for a claims adjuster? What are what are your employees coming to and saying, you know, “Gosh, if if I had less time to do this, I could really be there with my clients in-person more.”

Jeremy Gallant: Great question. So we’ve used this in various ways. I mean, I think our fast-track automation is one big one that, that that is out there. But we're also looking into, you know, how do we better utilize data for adjusters when they go out for a loss? So we collect a lot of engineering information and engineering reports on all of our locations. But wouldn't it be awesome if an adjuster could just say, hey, here's the location that I have had of loss; I press a button and boom, here is all the information I need relevant to get out of my claim. I don’t have to go look it up in a different system. I don’t have to print it out. I don’t have to sift through a 50-page document to find what's important. It just gives me what I need. It gives me how this location was underwritten. It gives me reinsurance. It gives me deductibles. It gives me everything in one spot. You'd be shocked how much time statutory filing of documents takes. So, another one is like, hey, how do we automate the filing of documents into the right spots? Sounds trivial, but, you know,  this is a lot of time spent on this. Capturing data. So, any input, you know, any notification of losses, we're trying to get them put into our portal, which will then help us facilitate the fast track or any other triage process we have, to be able to get the information to the right source. So, getting that data properly in. Now, on one side, a client could, manually input all of that data, but could we take that out from the email that it automatically populates into the system so that system will then start the automation process for smaller losses. We’ve been using or experimenting with Gen AI, looking at our operating requirements. So, if adjusters need a question, how do they go in and figure out what they need to do in a various scenario. It’s helping streamline lease and contract reviews. So, getting very quickly to the area of insurance of who's responsible for what, and is the policy going to cover what it is that they’re responsible for. We’re piloting using it for summarizing of documents. So, you know, in larger claims particularly, there's a lot of consultant reports that come back. These are big, lengthy documents. And they do come with exeitcutive summaries, but, AI can help us even summarize that even further to be able to get to the core areas. It doesn’t mean our adjusters shouldn't be able to read these things if they need to be. It needs to flag those items. But certainly that saves a ton of time of manual looking it up before. And just writing some of our summaries and file narratives that we have to do to document our files. We’re experimenting with AI getting us 90% of the way there and a human goes back to make sure, OK, these are all correct, but it’s able to extract that data from every other file that we’ve worked on in the claim to be able to summarize those nice and quickly.  

Brian Amaral: Thanks for being on the show.

Jeremy Gallant: Thank you very much.

Brian Amaral: Thanks for listening. Don't forget to like and subscribe. You can reach the show at [email protected]. See you next time.