Revolutionizing Risk Management with AI and Technology: A Conversation with CoreLogic's Garret Gray

19 min read
May 14, 2024

In Episode 4 of the Building Potential series, the President of CoreLogic Protect, Garret Gray, joins Archipelago's Founder & Chairman, Hemant Shah, to discuss:

  • How AI and technology are revolutionizing collaboration and efficiency in insurance claims, underwriting, and risk management.
  • The challenges and strategies in accurately assessing property values for insurance, including the impact of material and labor cost fluctuations on valuation accuracy.
  • Advancements in catastrophe modeling and property valuation at CoreLogic, with an emphasis on the importance of accurate data for effective risk management.

Watch, listen, or read along to the full episode below:

Episode 4 – Transcript

Hemant Shah: Hello. Welcome to episode four of Building Potential, the series in which we explore the frontiers of capability of the risk management and insurance ecosystem.

I'm Hemant Shah, your host, Executive Chairman of Archipelago, and I'm thrilled today to be joined in lovely Los Angeles with Garret Gray, President of CoreLogic Protect. Garret, welcome to the show.

Garret Gray: Thanks for having me.

Hemant: I really appreciate you carving out some time to chat with me about what's new in your world and what's important to the risk managers who interact with you and your capability.

Garret: Thanks for traveling all the way to LA. I mean, I guess we have some good weather to offer, but.

Hemant: Yeah. I just flew in from Chicago, so definitely an upgrade.

A few weeks ago, I had the opportunity to speak at your 2024 INTRCONNECT conference. Thank you for that opportunity.

Garret: Yeah. Thanks for that.

Hemant: It was a pretty impressive event with a huge turnout.

Garret: Thank you.

Hemant: It's one of the first couple of years where you've been integrating capability in one conference, right, across the insurance platform?

Garret: I think that's what's so exciting about this conference in general. I mean, besides the fact that I think we're taking a fresh approach to, you know, all insurance, including claims and underwriting risk management, bringing everyone together has been something that isn't normally done in our space, but has been received really well because there's so many connections between how claims are handled and how underwriting takes place and getting the people in the room to have the conversations and facilitating, away from the build bridges between each other, I think, is really important.

Hemant: Yeah. I noticed that. It was good to see the conversations going across the modeling community that I have some affinity for, the appraisal restoration community, the underwriting side. That was great to see that multidisciplinary.

Garret: Like, just the contractors. You just don't normally see contractors and carriers mixing like this. And, you know, we just think that there's such an opportunity to build bridges between all these constituents. Everyone has the same goal. You know, getting the balls over back to pre-loss condition and doing it in a way that, you know, keeps them happy, but also, you know, making sure there's affordable insurance that people can have access to. 

And in California, I'm one of those people who don't have a lot of access to insurance. And so, you know, there's big challenges that we're trying to tackle as an industry, and I think it's really needed to bring everyone together and see how they tackle those things as a group.

Hemant: So what were some of the key headlines that you were conveying at the conference?

Garret: Yeah. So we were focusing a lot on AI because AI is really starting to change our space in some pretty profound ways.

But I think beyond just it being a tech conference, it really is a facilitation of bringing people to rally around a problem that they can tackle in kind of a new way as a group. And so a lot of the content is designed to get people to think and then collaborate. And I think that's a, you know, from at least the feedback I've heard, hopefully, you know, you would agree, that there was a lot of that going on at this conference.

Hemant: Yeah. I look forward to participating. I guess, I'll be joining you in London in the fall.

So it's been it's been about two years since you unified leadership for CoreLogic's insurance data technology solutions. And you joined about a year prior by an acquisition of NextGear. Right?

What was it like being an entrepreneur, founding your own company, getting acquired, and then coming in to take on a much larger set of capability for the firm?

Garret: Look. You know, my whole journey has been quite interesting, at least to me. You know, I'm a college dropout.

I know it's sort of cliche.

Hemant: There's a lot of hacks in there.

Garret: I know.

But, you know, I started with this kind of really small focus problem on restoration contractors, you know, taking what I did out in the trenches, you know, as I when I was one, and turning it into a technology workflow solution that eventually then brought the carriers in to collaborate so that they could work, you know, more efficiently together to get to faster, fairer settlements. And, that was an interesting journey. Like, we had multiple acquisitions that we acquired, different, smaller companies and some companies are actually quite large. Stitch them together.

Those founders, believe it or not, are still with us today.

So we have had a really, you know, unique success of keeping the people that we've kind of gone on this journey with together.

Hemant: That's unusual. That must be a pretty compelling culture to keep a group of founders through roll-ups of still, you know, working collaboratively as a team under your leadership.

Garret: And a big part of my focus, you know, when you are a small group of people that are used to running in really agile and a way that doesn't have a lot of governance, because you're a small group, and then being consumed or acquired by a multibillion-dollar company, it's a whole different feel. And one of the things I'm really glad about is, you know, while there was probably a little bit of an interesting, rocky entry point because it's a lot of change. A lot of change and especially, personally, for me.

The kind of the ownership group and the CEO, that we currently have, Pat, has kinda recognized, "Hey. We want this group to run differently than the rest of ProLogic because they're playing a different game."

And, you know, we're in some ways a bit of a disruptor to what's happening in the space. And so with CoreLogic in general is the bigger player in the mortgage and tax and real estate side.

Hemant: More established.

Garret: More established and the market leader. We're the challenger brand, and so we have to keep that culture. So we we've kept a lot of our startup culture at CoreLogic.

And unifying these strategies, I think, has brought a kind of a refresh in how we're approaching these challenges in the insurance space, and we're doing it through a different lens, which I think is why you're seeing some momentum our way because carriers have been looking for some alternatives, and, now I think they found one.

Hemant: So one transition has been your own personal journey of entrepreneur, scrappy founder into a more established business

Garret: Yeah.

Hemant: And then taking responsibility for multiple product lines.

Have you been able to put your fingerprints on the new strategy yet? It's been only a couple of years, but it's been an interesting couple of years.

Garret: Yeah. I think so. I think, you know, and I don't wanna jump too far ahead, but even in the cat modeling world, cloud-based models are something that we've really focused on in the last couple years and delivered. So we've made some changes.

And I think there'll be more as we are bringing more AI solutions to our underwriting the cat-risk platforms. We're being very AI-forward, and so everything that we're doing is being impacted.

And that sort of mentality, I think, is creating some really fast change.

Hemant: Yeah. That was definitely a theme at the conference, stitching it together, technology and AI, across the whole platform of capability. So I think it's fair to say that a lot of people, in the risk management community, know the CoreLogic capability through the lens of the Marshall Swift appraisal and valuation suite of capability.

And I know that's where you're the gold standard for how property replacement costs, get re-appraised and revalued. I think it's fair to say as well that the insurance market, this has been a very hot topic over the past several years with underwriters pretty much across the board, flagging systemic concerns with their customers' property valuations, reevaluations, are they being adequately valued, and, you know, putting a lot of pressure on the upstream – the buyers of insurance – to take an honest and fresh and up to date view of this.

Some of the inflationary trends have moderated a bit in the last year. But from your vantage point what's been the drivers of some of the inflation trends? What's driving the change in the rate of growth? What are you seeing? Because you guys have access to a lot of data.

Garret: I mean, I think some of this is quite obvious in a way. Like, COVID changed the game. We had this really, kind of perfect storm of so many things that were kinda conversing at the same time. And then COVID, you know, disrupted the supply chains, and cost of materials skyrocketed, cost of labor skyrocketed, and it happened so fast that even though I think there were systemic problems before, in terms of, like, you know, in general, you know, most carriers were indexing, as opposed to refreshing, the data each year and recalculating the reconstruction value.

I think what happened, nobody would have seen coming. And so I think while it's a multilayered problem, there's probably lots of different ways to address it. The good news is is that it is starting to moderate. Like, for example, the cost of lumber has you know, there's been downward pressure on the cost of lumber, and now we think it's about to start to kinda bottom out.

Cost of labor is still high. And we think it's gonna be high all the way through 2024. It's sort of this double-edged sword because we think, obviously, if the labor market would get less tight and you'd have, you know, a little more unemployment, you'd see labor cost go down. But as long as the economy is doing good, which it, you know, still is, the cost of labor still keeps going up.

But we did not think that's what was gonna be the case. We were expecting, you know, a slowdown and that the cost level was gonna come down, but we're still seeing the rise.

Hemant: So interesting. So some countervailing trends. One is the materials costs, inflationary pressures are easing That's right. Part because of supply chain.

Garret: Supply chains have, yeah.

Hemant: They have reestablished their efficiency and so on.

But labor, which is a material cost of construction and reconstruction is still, pressure. I mean, I know when I talk to underwriters they continue to be very focused on this – as do the brokers working with their customers.

But often underwriters will scratch their heads and say, "Why is it so hard to get an accurate property replacement cost?" I mean, it's a core piece of information.

To some, it would seem to be readily available information. But if it was easy to do, this wouldn't be an issue. You know, it does seem like there's some – what are some of the challenges, you know, in getting a replacement cost valuation?

Garret: I also think they're trying to thread the needle of, like, hey. If we had a truly accurate – every year reevaluated, using all the property characteristics that we could, like, what would that do to premiums?

And so there's a retention issue as well. So I think, you know, underwriters are balancing, "We gotta retain our customers, and we also want to not have exposure that we're not prepared for." And so I do think that returning back to recalculating and bringing the latest data of what's going on with a particular property and getting what the actual reconstruction cost is, as opposed to just indexing it, will probably help carriers have a more accurate view of of their exposure.

Hemant: Yeah. A lot of the carriers' kind of vantage point on the valuation models are, I should say that some of the the general indices, index values by line of business or by region.

Garret: Right.

Hemant: Or, you know, many of them will use, I think you call it your 'express product', which is the three/four/five variable approximations.

Garret: Right.

Hemant: The cost per square footage for this property. But the real gold standard is having it done upstream at the source, right, where the owner of the asset actually retains an appraiser Yeah. To go collect detailed – How many variables power those models?

Garret: I don't know that I could even tell you.

Hemant: I think it's, like, dozens. I don't think of many underwriters, or some underwriters don't appreciate that to do a rigorous valuation is not a trivial exercise. It can include thirty, forty, fifty variables, describing the building. And it's a challenging modeling problem.

Garret: There’s ways CoreLogic can help with that because we know property better than anybody else. And, you know, for example, almost every house that transacts, it goes through our rails at some point.

Right. And so we have MLS data that we are actually combing through and going, okay. What's the quality of this kitchen? Is this a basic kitchen, or is this a high-grade kitchen?

And so we can actually, with a pretty, you know, a high degree of confidence, start to give those property characteristics that can help get a more accurate replacement cost or reconstruction value. And so it is a matter of, like, these things didn't exist, you know, five years ago, seven years ago, or even just a few years ago. And so I think carriers need to get more familiar with, you know, some of the technologies that are available that would actually get more property characteristics that would, does require the homeowner to do something, but maybe that is just data that's out there.

Hemant: Yeah. Because you're in a pretty, interesting position because, you know, at end of the day, rate is a function of, you know, exposure times – premium is a function of exposure times rate. There's there's basically two macro variables.

What's the rate, which is a function of hazard of the building, of the location and the volume of the property. And then what's the exposure?

So much attention in the insurance industry is spent on getting the rate right using cat models and rating engines and AI and predictive analytics. But if the exposure values are off by twenty, thirty percent, the rate's gonna be off by twenty, thirty percent. So getting this right.

But I have seen in my work, upstream with some of the large property owners and their broker partners. The last few years, there's been a real significant amount of attention played to getting not just the rate part of the equation right, but the exposure part.

And I think there's a growing awareness of the importance, not just we need to get our exposure valued correctly because the underwriters are asking for it.

We, as an owner, need to get our exposure, valued correctly because we need to know what our exposures are because we are ultimately the ones managing the risk of the asset. We wanna make sure it's adequately insured. We wanna make sure we have a handle on our aggregate exposures, and nobody benefits from a misvaluation of the property's replacement cost. It's not just for the benefit of the underwriter. It helps the owner and the manager of the asset do a better job understanding their own drivers of risk so they can mitigate their risk over time.

Garret: Well, think about, like, if you're a homeowner, like, your home is your biggest asset. Right? And I think before insurance became a more difficult thing to actually acquire in certain areas, like in California, there's a lot of insurers that have pulled out. There's so much wildfire risk. I live in an area that, you know, my CoreLogic wildfire risk score is high, which makes it harder for me to get –

Hemant: Yeah. So is mine.

Garret: Yeah. No matter what, I was just gonna say, no matter how many hints I've dropped, it just goes up every year. I don't know how that's happening.

But, yeah. No. I think this is a challenge of communication and making sure that there's an effective communication campaign to property owners that they wanna have accurate and a full value of what the replacement cost is. Because like you said, it doesn't serve anyone if if they're underinsured.

And so, obviously, that's gonna come with a premium adjustment. And so that communication plan is gonna have to be really thoughtful. But I think now is a great time to do it because everyone, you know, knows inflation's on the rise. So I think the gap between what they think or their expectations of things raising, and what the carrier will have to do to manage that is narrowing.

Hemant: Well, I look forward to continuing to work with you and your colleagues to develop strategies to get more accurate valuation driven upstream into the owners of particularly the commercial property owners who have trillions of dollars of assets and are increasingly, taking proactive measures to manage their own risk, not just by insurance, but to manage their risk and critical data. And accurate data is the foundation of all that.

One thing, I'm smiling as I say this, but when I was at – I've got to change gears to another set of your capability. At INTRCONNECT, it was a bit of deja vu attending some of the cat modeling sessions.

I'm a reformed catastrophe modeler, and for many years –

Garret: I have heard that you are, like, the founding father of cat modeling.

Hemant: Good thing the Zoom filters on set high. They just don't know how, gotta hide my age. But, yeah, I used to be a spirited competitor of your team and met several of my former compatriot competitors at your conference. But one thing that struck me was that there were a lot of sessions that you had where you were showcasing new capabilities.

Coming in from your background, with NextGear, coming into CoreLogic, how do you think about the cat modeling suite of capability that Protect offers, and how are you driving innovation or evolution of that capability? Particularly since you're up against some pretty entrenched couple of competitors we won't name that have pretty significant positions in the marketplace.

Garret: Yeah. Look. I think our focus is on really deepening our already superior science in a lot of ways.

As well as modernizing the infrastructure. Okay. So, there's a couple things at play. I think you know really intimately the difficulties of of getting to the cloud for cat modeling. Right? It's not an easy –

I already know. I have a lot of scars on my back. Yeah. And really tried to do that in the land of its time.

That's right. And so what's funny is when I was, on stage at INTRCONNECT, I don't know if you caught this, I said I'm about to unveil something that is gonna sound like it's not a big innovation, but it really is because we've brought our cat modeling and all of our 185 perils to the cloud. And everyone else in the room is probably thinking, we've been on the cloud for 15 years.

Like, what's the big deal? But this is hard, right? And people struggle to do it and to do it in a way that doesn't cause people to have to rerun all their numbers and they aren't getting consistent outputs back.

And so what was one of our major focuses is, like, if we're gonna move this to the cloud, people have to rely that what the answers they got on prem solutions are gonna be the same as what they got in the cloud.

And so we were able to achieve that, on top of also speeding up the process by orders of magnitude, or things that used to take, you know, days to run or running in hours. And so it's it's something that is really starting to speed up the process of getting information back to the carrier.

So on top of that, we're looking at or we've already delivered for them to be able to access this through APIs or through other distribution networks or platforms. And then, obviously, we've got the cloud as well. So there's meeting customers where they wanna be and how they wanna consume information. And that's a really big focus for how we're modernizing the infrastructure.

Hemant: Yeah. For years, the principal dimension of competition amongst the modelers has been on the science, and there's a great intellectual and methodological capability in all the modeling firms. But, you know, increasingly, as the industry has operationalized their workflows around using these models deeply into underwriting, rating, portfolio management practices, the technology dimension is critical as well. Like, how are these models delivered, how they package, and, you know, moving into the cloud with an API driven component-based approach that lets your customers embed that IP into their workflows in a more efficient way is important and a smart play.

Garret: Yeah. I think it's one way we can be more competitive. And we realize we're not always gonna be the, you know, the choice for certain carriers. But if they want a secondary view of risk and it's easy to work with us, there's there's ways for them to, you know, consume that with us.

Hemant: So as any, all experienced modelers know – look at the camera. All models are long. Some are useful.

You know, it's important that the industry understands this, that, the models have uncertainty. There's methodological debates. There's sometimes unlimited data, yet the models are useful. And when they get embedded into workflows, you can make better decisions. So that was, yeah. I definitely picked up that at INTRCONNECT.

So what are some other main headlines for 2024 coming into your modeling capabilities?

Garret: So, we're really proud of our severe convective storm model.

One, it's been adopted by a lot of the top ten US carriers. So, we were happy with the adoption of it. But, also, we have pretty significant weather forensic capabilities as well.

Hemant: Interesting. That's part of your cross-platform capabilities.

Garret: Yeah. It is, so this is where claims and underwriting and risk management can kinda come together.

And so our science is being informed by what's actually happening with our forensics, which I think is a differentiator for us and one that I think will give, a lot more accuracy to the process.

Hemant: Yeah. That, Garret, that really resonates. I mean, one, you know, in the not too distant future, perils like severe convective storm, tornado, hail storm, straight wind were kinda relegated to secondary perils.

And that term secondary perils kind of seeped its way to the market as in they're not as important. But the reality is if you look at the loss experiences for a couple of years, it's being driven by not the tail events, but the body events. Yeah.

There's still major catastrophes, but the frequency and the aggregations of these, severe convective storms, are driving a lot of the actual loss experience of the insurance industry and, arguably, more importantly, are driving the actual loss experience of the property owners and the commercial entities who are trying to manage their risk and buy insurance. And so, focusing on severe convective storm - and, you know, I know many in the insurance industry have always been skeptical and not entirely satisfied with the rigor of the severe convective storm modeling as it compares to, say, the primary peak perils that the modelers grew up modeling, earthquake, hurricane, etcetera.

So I think, focusing on that makes makes a lot of sense from my mid vantage point.

Garret: I think there's a really interesting interplay with claims here, because I think one of the problems is that when there's a hail event that comes through, one of the things that happens in this industry on the claim side is roofers know from the news - okay, hail went through this area. They don't necessarily know where the damaging hail fell, but they just blanket in the area and go, "Hey, we might need to replace your roof." And they have this army of sales folks that are out there knocking on, Paul Solarz' doors. And, you know, if you're a Paul Solarz, you've paid your premiums for thirty years and you're like, "Oh, can I get a new roof and let the insurance company pay for it?" There's kind of a – not necessarily an intentional – but an opportunity for claims to be paid that shouldn't be paid.

And one of the things that we can do with our weather forensics is we can actually look at, okay, when this hailstorm kicked through, were you actually in the path of damaging hail? And at the moment of first notice of loss when the home run calls in, we can serve that up to a carrier so they can say, "Hey, actually, this isn't... we don't think any hail actually fell on your house that would be damaging enough."

And now they're kinda just paying it. And so the frequency is even higher than it should be because there is just this perception, "Hey. There's hail in the area, so it's time to get a new roof." And I think we can combat that.

Hemant: Yeah. Because the capabilities you offer are clearly not just the prior cat modeling itself. It's the forensics. And then it's the actual assessment of the damage.

Garret: And all that feeds back into our models.

Hemant: A full life cycle. That's you.

Garret: That's right.

Hemant: So I can see how your platform, you know, broadly speaking, can curate, a set of differentiated capability across, not like in the modeling, but the understanding of the actual damage patterns, claims, information, the modalities of damage and feeding that back into to the models. Well, here we are, we're just about in May, so we'll see how what 2024 brings.

We have the severe convective storm, climate hazards, and general hurricane. You know, the market has been – it's been, what, seven years of a hard market? And we'll see. But I think it's important, hard market, soft market, to have, you know, sound capability to value exposures, measure risk, and more importantly, by doing so come up with effective strategies to mitigate and reduce risk in the first place.

So I've enjoyed, chatting with you, Garret, getting to know you better. Again, thanks thanks again for including me in your conference. Look forward to seeing you.

Garret: See you in London? Yeah.

Hemant: See you in London again. And thanks for giving me the opportunity to come to LA and chat with you in person in this lovely town.

Garret: It's been amazing. Thank you.

Hemant: Okay. Thanks, Garret.

Get Email Notifications