Products

Solutions

Resources

Get a Demo

Faster claims, better decisions: Balancing AI speed with human judgment

Nov 2026

Dave Tobias and Sean Serafin



Sean Serafin and Dave Tobias explore how AI is reshaping claims, where human judgment matters, and why clean data, strong processes, and quality must come before scale.
Photos of Dave Tobias and Sean Serafin for episode 5 of Nearmap podcast

“The insurance industry is data rich and insight poor. Now with AI, that’s no longer the case. We now have the ability at scale to take massive amounts of data that we could never, even unstructured data, and make meaningful use of it.”

Sean SerafinFormer Chief Claims Officer and SVP, USAA


Episode 5 transcript

[00:00:00] Dave: Hello, and welcome to the Reality Layer podcast. I’m Dave Tobias, Chief Product Officer at Nearmap. Let’s get into today’s episode. Today, I’m joined by Sean Serafin, former Chief Claims Officer and Senior Vice President at USAA. Sean has spent more than 25 years leading claims operations and transformation, with responsibility spanning property, injury, catastrophe response, investigations, and recovery. In this episode, we will explore how claims leadership is changing, what it takes to modernise operations at scale, and how carriers can use technology and automation without losing the human judgment, empathy, and trust that matter the most when somebody’s having some of the worst days of their lives. Sean, welcome to the podcast.
 [00:00:45] Sean: Thanks, Dave. It’s an honor to be here, and I look forward to talking with you.
 [00:00:48] Dave: Sean, you spent more than 25 years in claims across everything from investigations to property, cat claims and response, enterprise claims leadership. What are the major innovations you’ve seen in this space? And what do you predict to be the next big one?
[00:01:02] Sean: Well, it’s been an incredibly interesting run in our world. And, you know, it started years ago, right, with just moving off of paper files. And I think it’s really interesting because some of what we’re seeing now is similar in terms of we thought we were going to go digital and it had all these great benefits. And it did, of course. But it also in some ways slowed us down at first. And I think we’re in that same loop now as we innovate with AI, but you look at the history, you look at all the things we had that evolved innovatively. We had aerial imagery. We had connected data in vehicles, right? These things are all things that have changed our industry. And then you look at fraud detection and the automation that happened there and the insights. Then you start looking at machine learning. Machine learning and automated process was the next big horizon, and that helped us be more consistent, more accurate, helped us with speed. So, these things all have layered up to where we sit today with obviously AI and the excitement that that brings, but also the concern, right? There’s a lot that AI is going to do for us, taking that automation and those processes I mentioned to the complete next level. And so, I don’t know if I’m the best guy to predict what’s absolutely going to come next, but I will say there’s a couple of things I think that jump out to me. The first, and we’re going to talk a lot about this, I think, today, is there’s going to be this positioning of AI to AI, right? We’re going to get to a place in the next several years where a lot of interactions, not just in insurance, but across our entire ecosystem, are going to involve AI agents and voice agents and chatbot agents interacting with each other and resolving things without any human interaction. But with that, there’s going to be points in our business that require human intelligence, human decision making, and that human in the loop, human on the loop. So, I think we’ll talk more about that today. But I think that agent-to-agent interaction is going to be the next game changer in insurance. The other part I would tie to it is just training. We used virtual reality training years ago, maybe seven, eight years ago. We started cycling that up. It was super powerful way to teach people. Well, now we’re going to have that with AI and all kinds of new layers to help teach those people that we need involved. And they’re going to have to learn different skills. So, I think the two big innovations we’re going to see is that agent-to-agent interaction and then what that means as far as people and how we interact as leaders and then how we recruit talent and how all that comes together to deliver excellence for those homeowners and those insurance policyholders.
[00:03:34] Dave: No, good, good viewpoint, Sean. And I happen to have lived through some of this agent-to-agent interaction over the last two weeks here. Unfortunately, I had some pet insurance claims, and I decided to use my OpenAI Agent to file the claims and the paperwork for me. So, I’d never filled out any forms myself. I had the agent do it all, and there was multiple claims, and it was kind of complicated. I never touched the claim except to submit it, check the information, click submit at the end. So I don’t just think we’re going there. I think we’re going there quicker than we all thought.
[00:04:09] Sean: That’s exactly it. And the speed of it and the availability of it is just growing exponentially. So, we’ll see a lot more of that. And everything from buying policies to settling claims, the whole thing. So more to come.
[00:04:21] Dave: So, it seems like claims leaders today are under this constant pressure to improve speed, consistency, efficiency, customer experience, do all these things all at the same time. How have you seen that job of leading a big claims organisation change over the last decade?
[00:04:39] Sean: It has changed a few ways. And I think one of the earliest is just in philosophy, right? I mean, I think it was probably for some time viewed as a necessary cost centre. Then it kind of moved to a commodity. And then now I think it’s viewed as a potential deep advantage for the carriers to understand it, right. And if you think about any of the pressures you just talked about, any carrier that’s bringing in the revenue, there’s the bulk of that revenue that goes out through the claims process as it should. That’s by design. And so the pressure to be accurate and be fast, but also deliver empathy and experience is bigger than ever. And so, you know, I think when you look at that and you think about what the challenge has created there because of those pressures, it creates a place where I think the leaders have to think very differently. They first have to do what we just talked about. They have to be able to take advantage of technology and AI and these tools that are available. They can’t be slow. They have to embrace them. But they also have to be able to balance that really, really carefully with the people side and with the empathy both for employees and for the customer. Right? And so I think it takes a different skill set to be able to do that. You can come in and you can drive hard results and, you know, you can focus on one element. But to focus on all the elements and manage that bubble or that triangle, that’s a whole different skill. And so it’s going to require leaders to up their game for sure.
[00:06:03] Dave: So as more of these claims become automated in different ways, right? Whether it be agent-to-agent or other automation that we’ve brought forward, what skills do you think will matter most to the next generation of adjusters?
[00:06:16] Sean: Yeah, it’s interesting because I think critical decision making is going to be one of the top skills that are going to be the thing we have to look for, the thing we have to make sure we reinforce. And I think that’s a tougher thing to find than it sounds. So AI is going to do a lot of things and it can actually get to a place where it’s going to make decisions and really good decisions, but you’re still going to need that critical thinking layer on top of that. And so I think if I was outlining just two key skills and talents, it would be tripling down on critical decision making, and then strategic thinking. The gap that I see the most in our industry, Dave, is even at senior levels, is a challenge to be able to think strategically, to look ahead and to think about what’s coming and then build backwards three, four, two, five years, whatever the window is, and be ready for those changes and have disciplined plans that you can execute and that get the job done. That’s an ongoing challenge. And so I think strategic capability is going to be more and more important than ever, especially as you’re doing what we said. Integrating technology, protecting these results that have to be disciplined and correct, but also protecting empathy and service and all of those things that go back to the policyholder every time there’s one of these interactions. So those two, I think, are two of the biggest skills that have to be sustained.
[00:07:37] Dave: There’s a lot of excitement around automation, around AI, but claims often involve judgment calls that are highly contextual. Where do you think the technology should take work off people’s plates? And where do you think human judgment should still come in and be firmly in the loop?
[00:07:54] Sean: Yeah, there’s two places. I mean, I think we sort of coined this idea that speed and ease can become empathy, right? So in other words, if you make something easy and you don’t make me do a lot of effort on my own, especially as a policyholder, that translates to a level of empathy. But there’s a whole other part, right, that you and I know, Dave. If your house is flooded and you can’t use your kitchen and you’ve got kids and you have to relocate to a temporary property, I don’t care how quick it is. I don’t care how great the digital interactions are. There’s a point there where you need somebody to wrap their arms around you and say, we got you. And this is what this looks like and to communicate clearly and transparently. And that’s the place I think to your point, your question, where we’re going to continue to need that really precise human interaction and what I said earlier, critical thinking, because juggling those trades, juggling the repairs, juggling the ins and outs of putting things in storage and all those pieces is a lot of effort and it’s a lot of pain, no matter how you try to do it for the member. And I don’t think those are things that go away magically with AI. You’re still going to have to do those repairs. You’re still going to have to put things in storage. So I think those are areas where you’ve got to have that critical thinking. You’ve got to have that care. And those are going to be precise touch points when you’re in those repair process elements that matter the most. The rest we can take off the plate, right? Subrogation, things like that. There’s all kinds of administrative work that we can speed up and enhance and make people more effective at. But those critical touch points where you’re interacting with the loss are still going to require that human interaction and that touch point.
[00:09:28] Dave: So put your futurist hat on for a second here. What do you think will absolutely still require a human in the loop five, 10 years from now, no matter how good the technology gets?
[00:09:40] Sean: Yeah, I think it’s going to be a couple places. I think, as you automate more and more, you need a lot more oversight of that automation because the worst case right is you automate bad process and you do it at scale and now you have really bad outcomes and you don’t have a bunch of people sitting there to pick that work back up if you’ve changed the system. So as we automate more, I think you have to have really smart people sitting on top of that to manage it and monitor it and make sure it’s performing the way you think. And then the second place is what I just mentioned a little bit ago, right? Just I think you’re going to need people that interact at key moments that matter when that policyholder is going through a traumatic loss. And there’s going to be several touch points throughout that process that just require a human interaction, a person in the loop, not only making decisions, but also delivering that service element that is still tough to deliver digitally, even though it’s gotten better and we can do things around it. You need that human touch point. So I think there’s those kind of bookends, right? It’s through that process of the repair and managing that repair. And it’s managing all the automation that we’re going to have in place. You need both.
[00:10:43] Dave: So you’ve been on the ground during some massive catastrophe events. What’s something that you simply can’t learn from just sitting in HQ looking at a dashboard that you have to really be on the ground for to see and understand?
[00:10:56] Sean: Yeah, the part, Dave, you can’t replicate being in an office. And I always encourage people, if you’re with a carrier or a TPA or anybody and you have a chance to deploy, do it. It’s so enlightening. But the thing you can’t learn from your seat without being out there on the ground is really the impact to people. I just, you can see it on TV. You can watch it. But when you walk the ground and you understand what it’s like to have no power and not just no power, but you can’t find ice for 40 miles. You don’t have any food or restaurants available. Grocery stores are shut down. Roads are not able to be traveled. When you have to live like that and live like that for weeks, maybe months, but for weeks at least while the emergency crews get in, the toll it takes on people is just immeasurable. And you go and see that. And I think it has to change your mindset. It has to change your empathy. It has to change how you build process, it has to change things like we’re talking about now. Where exactly do you deploy AI? And where is that probably not the best place to deploy AI? So, I mean, I just think the impact of people and how that has to change your mindset, everything you do end-to-end is so powerful.
[00:12:03] Dave: So, you’ve had the opportunity to lead some big transformations and some very large structures. Where do you see these programs most often get stuck? Is it the technology, the process around it, or is it the people and the reluctance of the people? Where do you see that?
[00:12:20] Sean: Yeah, I mean, I think all those are things, right? But as I look at what I’m seeing now, especially, so both prior transformations, but what I’m seeing specifically with AI, I think there’s a few places that you see it get stuck. And it’s really in back, what I would call back-end preparation. I think it’s really, if I was taking three things away, I would say you’ve really got to get into, is your data mapped? Is your data clean? Is your data accurate? Because if you’re trying to process AI and you have a prompt that’s running and it’s pulling all these data sources to determine. I’ll use an example, say coverage of a claim, the policy status, payment status, all these elements, and then provide some response back to somebody, whether it’s an adjuster or a policy holder. All that data has got to be tight. And then the second part is, have you mapped the process? So, do you really have good process? Because if you don’t, you’re just going to layer AI on top and you’re going to run with a maybe a poor process. And like I said earlier, at scale, and you’re just going to have massive issues that are very difficult to unwind. And then the third one really is, are you really invested in a strategy? Too often right now, what I’m seeing is you got everybody from CEOs to frontline folks that are talking about AI and they want AI and they want to check a box and they want to feel good, like they’re not falling behind and like they’re engaged in it. And so, what you’re seeing is a lot of deployments that are that are really fractured. They’re point solutions or they’re fixing a certain problem. And in some cases, it’s not bad because you’re learning and you’re getting educated. But if you aren’t careful with that, you end up with a really fractured deployment. What you see is people chasing expense and then they’re upset because they’re not getting the expense they expected, right? So they’re looking at it and they’re going, oh, we didn’t recoup expense. Well, it’s because of those three things. It’s poor data. It’s poor process and it’s not being strategic. And so you’ll get some very small lift and you kind of feel good and check a box. But those are the things I’d say, like people have to be really, really cautious as they deploy. Make sure they’ve covered those three things.
[00:14:20] Dave: So with AI specifically, there’s this push, as you mentioned, to get it into processes, right? To get it out there, from the CEO down at many of these organisations. And there’s a lot of failed POCs happening because of that. Maybe the people aren’t ready. Maybe the process isn’t ready. Certain cases, the technology is not ready. But put you on the spot as somebody who’s sat in the seat of a chief claims officer. If you were going to try to pick one metric, and I know this will be hard, but pick one metric to track and look at every day for one of these implementations, what would it be?
[00:14:57] Sean: Yeah, I don’t know. I can say that there’s one metric I would track, but I think if I was really looking at this and pursuing it, and we touched on this earlier, right? A bulk of what you do is indemnity and a bulk of what you are doing is the money you spend on the claim itself. So I think if I’m looking at one overall metric and you box me in, I’m going to say it’s quality. It’s got to be quality because, and I’m going to cheat a little bit and say it’s the quality of your program. So the quality of your deployment, your AI, the accuracy, the data, all those pieces, all that accuracy, but also the accuracy of the output. Are you driving accuracy in your ultimate claims result? Because if you are, that trumps everything. And that gets us back to affordability and availability in the marketplace, which are huge challenges right now across the industry, right? So if you can get accuracy right. You’re compliant, you’re safe, and you’re, you know, it’s better for our entire ecosystem because we can do better with availability and pricing and affordability.
[00:15:56] Dave: So you’ve had the opportunity to both be on the digital strategy side, but also the operational side. How do you tell if a new piece of technology is actually improving claims performance or just adding another layer, another form an adjuster has to fill out, or complexity, even if on the surface it seemed like it was going to be a good idea?
[00:16:15] Sean: Yeah, to get that right, you have to approach it from multiple angles, I think. And so, you’ve got to do a few things. First, you should have a very disciplined upfront expectation, right? You should have documented what are you at today and what are you expecting to see after you’ve deployed this thing, right? Whatever it is. And so, I think the first key part is map that out and have really disciplined metrics so that you can measure whether it’s actually improving the bottom-line result. And, like I said earlier, that way you’re not chasing some expense, for example. And we kind of always joke, right? Like you got, well, you got a finger here and an elbow there and a knee, but you didn’t get a whole body. So you didn’t really have any FTE or, you know, headcount reduction, which I, by the way, I don’t think should be your first target anyway. But, but you don’t have that, right? So disciplined measurements, number one. And with that behind that, I think you got to look at things like adoption. You got to look at throughput. You’ve got to look at, you’ve got to do some employee pulse results and hear from your front line. Is it making their job better? Are they more efficient, more effective? And do they feel better like they’re serving their customer, their policyholder better? So you have to really come around it from all those angles and say, was there disciplined metrics? You got to look at the quality I mentioned earlier. And then you’ve got to look at the engagement of the employees because they will tell you if it’s making their world better, then I can guarantee you the world’s better for the customer. If the employee world is not good, then the employee is not going to translate a great experience onto the customer. So you got to look at that whole circle and that 360 view.
[00:17:45] Dave: Yeah. I told you it’d be hard. One metric to rule them all is always a tough one, right? But claims teams have access to more data than they’ve ever had, right? There’s a wealth of data out there, both pre-claim on the underwriting side and now post-claim and during the claim, right? There’s data flowing in constantly. How should leaders think about incorporating all this data into these claims processes? We have AI that can help make sense of it, but what’s some advice you give to some of the folks sitting in that seat now, trying to make sense of all the data coming in? What’s some advice on where to start?
[00:18:22] Sean: Well, it comes back to our conversation, Dave, about critical thinking, right? And understanding the things that matter. And so I would say don’t be enamored with shiny objects, right? Focus on those few things. There’s only a handful that really make a difference. Focus on those and get after it. And here’s the beauty. We have said for years that the insurance industry is data rich and insight poor. And you just said it. Now with AI, that’s no longer the case. We now have the ability at scale to take massive amounts of data that we could never, even unstructured data, right? That we couldn’t take before and make meaningful use of. Now we can. So, yeah, to your point, don’t overwhelm your people with that. Be really specific and then give them and push to them some really, really powerful insights, but narrowed in scope and with what I said earlier, right? With clean data, make sure your data is clean first and then push it to them. But really, you got to shrink it and package it into those things that matter. There’s a handful of metrics in our business that make or break everything. And so, focus on those. Get those right, get those really strong. And yeah, don’t overwhelm your people because it’s they’ll be you know in analysis paralysis if you do that.
[00:19:29] Dave: So, you mentioned it before, but you’ve sat on both sides where you had the responsibility for member experience, but also employee experience. When those two were aligned, when that was right, when that through line really, really performed, what did the insured see? Or the claimant, how did that experience change? When those two were synced up, what was the benefit?
[00:19:52] Sean: Yeah. And I think, by the way, the most successful carriers are going to sync those two up. And it’s hard. It’s a lot of work to get both those right. Right. And it’s not easy. But here’s what the customer sees. They see clear, concise, transparent communication. They see outcomes that make sense to them and they don’t have to put a lot of effort in to get there. I mean, that’s essentially what they see. It’s, in other words, it’s a clear, transparent process, doesn’t take a lot of work, and they don’t feel like they have to dispute what the outcome is because they understand how you got there. So I think in a really clean world, that’s what you see. And that just gets faster and shrinks with our new world that we’re in with the technology and the automation we have now. But ultimately, that’s still what they should feel.
[00:20:39] Dave: Yeah, you said it. I mean, done, done right with the AI tools that are available, you should be able to deliver on all those vectors, right? More transparency, more communication. Ease of workflow, speed, right? It all should be there because, you know, the tools can help you unpack a policy, unpack, you know, claims exclusions, et cetera, et cetera, right? But you have to use the tools in the right way at the right time. And I think that’s a really important, you know, element. It’s just because you have a hammer, everything shouldn’t be a nail when it comes to AI. So, you’ve had over 25 years in claims leadership. What’s one piece of advice you find yourself giving people who want to lead in high pressure, high accountability roles like the ones you’ve sat in?
[00:21:22] Sean: Yeah, I think there’s two things that jump out at me, Dave, and as I look back on my career and think about those questions, right? And the first I would say is be uncomfortable. Make yourself uncomfortable. The times that I was uncomfortable and I was in challenging roles or I took something on I wasn’t sure I was ready for, you know, we could spend an hour talking about imposter syndrome, right? That’s where I learned the most. It’s where I grew the most and I gained the most skill and my viewpoint opened up. So I’d say, you know, make yourself uncomfortable and do it early. Don’t sit in a comfortable place for too long. You know, so that’s one. The other one I would say is. Everything is relationships. Get to know people, understand people. You have to know your peer group. You have to know what motivates them, what they’re thinking about, because the only way things get done is when good people come together and make it happen. And so you need those relationships to make that occur. And so those two things I think will get you through every level, whether you’re starting out as a leader or whether you’re a senior vice president running a huge operation. It’s the same fundamentals and those two matter.
[00:22:24] Dave: Sean, as you look back at your career, what’s one thing from a transformation or technology perspective that you implemented, and it went well with success in your mind? And what’s the opposite of that? Something that you tried and it just didn’t quite work.
[00:22:38] Sean: Yeah, interestingly, I can use the same example in two ways. And I think it’s just an illustration of what you’re asking. So I mentioned earlier using virtual reality to do training. And so I deployed that at a company and it worked amazing. I mean, it was super engaging and we had just demonstrated results and how our quality went up. People felt we centred this around things like difficult resolution of claims, right? Water losses, things like that. People felt more confident. They felt like they were better equipped. We gamified it, right? So, there was some competitive piece to it. And so tangible improvements in quality and outcomes and engagement. Tried to replicate that at another situation, another company, and it didn’t go anywhere near as well. And it was interesting because you would have thought that we kind of mirrored the first thing and how come it didn’t go well the second time. But it really came down to how we implemented it, the culture, lack of buy-in. And then it was just kind of like mid-COVID, right? So, you had these elements of people are sitting at home and you’ve got to be shipping them stuff. And by the way, they’re on their computer all day on Zoom. So, I think it’s just a good illustration, Dave, of. Even though you have the technology that can work here, it can work there, you’ve got to still pay attention to every one of those elements. All the things I mentioned earlier, the process, right? The employee engagement and adoption. Have you set it up right? Have you got buy-in into it? And are you articulating why it’s going to make a difference and how it’s going to be better? I think we didn’t do that as well that second time. And we just didn’t get adoption for it. And it fizzled on the vine, which was unfortunate because I think it’s a good product, but it was our fault. It was the way we deployed it. So I certainly learned from that and learned how to, you know, you can’t pick it up and lift and shift. Sometimes you got to start from scratch and really rethink it based on your environment, your culture, all those pieces. So sometimes it works, sometimes it doesn’t work second time.
[00:24:32] Dave: No, good story, good advice. Sean, thanks for spending time on the podcast with us today, bringing all your claims experience to all of us here and really appreciate it.
[00:24:43] Sean: My pleasure. It’s been great to talk with you, Dave. And I look forward to the next podcast.


Dave Tobias

Dave is Chief Product Officer at Nearmap, where he leads product strategy across the property intelligence portfolio. He co-founded Betterview, the application now used by carriers across the US for underwriting and risk selection, and joined Nearmap through its acquisition in December 2023. He has spent more than a decade building the tools insurers use to understand property risk.



Sean Serafin photo
Sean Serafin photo

Sean Serafin

Sean Serafin is a claims executive with more than 25 years of experience leading large-scale operations, transformation, digital strategy and technology implementation across the insurance industry.

Most recently, he was Chief Claims Officer and Senior Vice President at USAA, leading claims operations including injury, property, auto physical damage, catastrophe response, contact centre operations, investigations and recovery. He also served as Deputy Chief Claims Officer, VP of Property Claims, Head of Property Operations and CAT, Head of Claims Central Services, and AVP of Claims Security.



See the Reality Layer

Every episode covers claims, catastrophe response, and the AI reshaping insurance technology.

View more episodes

Explore the latest

picture of Nearmap Labs

Nearmap Labs

Explore the research and early-stage technology behind what's next in property intelligence. Get hands-on with new tools before general release, and meet the people building them.

Explore Nearmap Labs
picture of Customer Stories

Customer Stories

See how carriers put property intelligence to work across underwriting, renewals, claims, and catastrophe response, and what changed once they did.

Read Customer Stories
picture of Webinars

Webinars

Go deeper on the topics the show covers. Sessions with the practitioners and product teams closest to the work, available on demand.

View Webinars