Beyond a single view of risk: why catastrophe modelling is becoming more collaborative

Matthew: [00:00:00] Great to have you. We’ll talk a bit about this, but we’ve had a very similar career for many years, so look forward to hearing a little bit more about what Verisk is up to. We’ve got a heat wave here in the UK, so apologies if I’m looking a bit hot, but are you calling in from Boston? 

Rob: I’m calling in from Boston. Not a heat wave here. In American temperature, it’s about 70 degrees Fahrenheit, so very temperate for this time of year. 

Matthew: Yeah. We all complain after about three days of heat, so we’ll be looking forward to the cooling down in the UK later on. So you are President, Catastrophe and Risk Modelling at Verisk. And so Verisk built its business around being primarily a modelling company. Now you’re opening up to have a universal exchange. As I understand, you’re allowing other models to come onto it. What’s responsible for that shift in strategy for the business? 

Rob: Yeah. Honestly, Matthew, I don’t know if it’s a shift in strategy. This is something I think we’ve been building to [00:01:00] for the entirety of our nearly 40-year career as a catastrophe modelling firm. We’ve always been very open, starting from open exposure data formats. And actually, this is our second foray into this field. We had an application within our Touchstone platform called Model Builder that for a variety of reasons never really took flight, but always with the aim of providing access to other views of risk, and importantly, giving our clients the opportunity to deploy their own views of risk or their own models within one catastrophe modelling platform. 

Kudos to Simplitium and the NASDAQ platform. They cracked the nut on how to operationalise the Oasis loss modelling framework. And that was the real impetus to get a lot of these providers deployed onto one platform that opened up different views of risk to the market. We were on a partnership path with NASDAQ, had the opportunity to acquire the platform, and fortunately were successful in doing that. 

So it’s arguably a continuation of our strategy of giving choice and opportunity for our clients and [00:02:00] prospects to truly understand global catastrophe risk. One piece of which is Verisk, and one piece of which admittedly isn’t. 

Matthew: Yeah, really interested to hear that news about Verisk’s acquisition of what was previously the NASDAQ model and enabling the Oasis platform and models in there. Kudos to Dickie Whitaker for initiating a lot of those smaller models and giving the access to people. And also, it will come into a little bit when we talk about that global protection gap. But just from your own background, you started off in that world of insurance-linked securities, and these days we talk about parametric insurance. But for you personally, what’s been your development from that area of the business into your role today? 

Rob: Even before that, Matthew, my very first job was at AIR back in 2002 as an entry-level risk analyst, taking client exposure data, processing it, putting into the models and getting output into their hands so that they can decision it through different parts of their workflow. I was fortunate to have the opportunity to step in and help grow the insurance-linked security practice really at a time when the market exploded. And I was lucky enough to be [00:03:00] able to build a team that grew that space from a modelling perspective and continues to have a good share there now. When you’re front line like that, you’re deploying models, you’re talking to clients, the nice thing about the catastrophe bond market is you, because they’re so high profile, you have the opportunity to engage with company CEOs, company presidents, company chief risk officers, these individuals who are really on the front lines of protecting their organisations. And that gave me the opportunity to then step into different roles in Verisk, leveraging that experience and that knowledge from talking to those people. So been lucky to lead our global sales function. I looked after marketing for a bit, client service, operations, IT. Really anything in this business that was not actually building the model or coding software I was lucky to have the opportunity to do. And then when my predecessor decided to step away, they gave me the keys to the ship, so to speak, for a bit, and it’s been really fun. Certainly a lot of learning and a lot of really great experience. 

Matthew: So can we just talk a little bit about Model Exchange? What problems does this solve for your clients, and how are they using it? 

Rob: The biggest problem I think it solves is [00:04:00] gaps. No model provider provides a model for every single peril in every single region around the world, and there simply comes a time when the risk is evolving at a pace that’s a little faster maybe than we can integrate it into a global catastrophe loss modelling platform. So flood is a good example where there’s flood risk in many countries around the world where currently Verisk does not have an offering. So the benefit of Model Exchange right out of the box is these providers who have these views of risk that we don’t have, it enables our clients or our prospects or the market to get a view of risk to potentially sit alongside a Verisk model for another region. So if you’re truly writing a global portfolio, this would give you the opportunity to dive in and have that comprehensive view of risk together. It also will provide a framework where clients can ultimately deploy their own views of risk. So I mentioned the Model Builder platform as the predecessor to what Model Exchange will ultimately deploy. But think of it as just a framework where you can put hazard, vulnerability and a loss model together to [00:05:00] get a view of risk. Whether that’s a third party, whether that’s a Verisk view, or whether that’s your own internal view, Model Exchange provides that workbench that lets you have a model framework deployed.   

Matthew: And any platform to be successful like this and open it up to third-party models needs to have companies that are willing to work with you. Who’s actually putting their models on the platform, and how easy or difficult is it for someone to do that? 

Rob: I want to amplify your comment of the great work that Oasis has done for years. It was difficult for quite some time to get a model deployed onto Oasis, and the real problem that NASDAQ solved was making that much easier, and they’ve continued to develop and deploy a process where third-party model providers can engage and get their model deployed much more easily than they could before. We have nearly 20 providers, almost 400 models deployed globally. We put out some press yesterday that the latest company to engage in Model Exchange is KatRisk, and we’re really excited to have them on board. So it’s pretty difficult to find a region around the world where there’s not a model or a view of risk deployed on the Model Exchange platform, [00:06:00] and we’ll continue to grow. There’s no one that we prohibit from deploying on the platform. We’ve made overtures to, as far as we know, every modelling company in the world to deploy their models on the platform, and hopefully we can get to a point where we have that true universal global coverage.   

Matthew: I’d missed the news about KatRisk, but that’s great news. They’ve been making some fantastic developments. Known the team there for a long time. Of course, if you can be that platform and intermediary between the data and the analytics and the end clients, clearly it’s good for you across so many different areas, not just in the world of catastrophe modelling. We’ve seen companies that have either succeeded or failed depending on how well they can collaborate with third parties because it’s just so difficult, as we all know, to get into an insurance company, get credibility, go through the buying cycle. So I would suspect for your partners, that’s also a big attraction of this, is that they themselves are getting presented directly to your clients, and therefore they have an opportunity to sell their products.   

Rob: Exactly right. It opens up scale for them. So we’re in a lot of rooms that maybe historically they have not been in. It brings clients, it [00:07:00] brings opportunity, and ultimately I think we’ll get here in our conversation, but it really is about providing solutions to help make the world more resilient to natural disasters, preparing for, responding to these absolutely tragic events. And if there happens to be a model in a part of the world where even if we have a view of risk, if there’s a side-by-side view or there’s an alternative view that we can provide that helps a client be better prepared, then we’re happy to make that introduction and happy to sit side by side with these other model providers. So for them it really is about opportunity and how can they get into places or companies maybe where it’d be more difficult for them to get just because they’re usually pretty small. Their staffs are usually pretty small. 

Matthew: Yeah. We’ll definitely get to that resilience piece, but just a couple more pieces on the models itself. So you’ve got a very strong brand name, rightly, a long history, Verisk and AIR before that. How do you think about the companies that come on? Do they have to go through some kind of validation because it’d be either implicitly or explicitly you’re giving some kind of endorsement, I suspect, to some of your clients if they’re appearing on your platform? 

Rob: I think importantly, if we’re gonna have a platform, [00:08:00] Matthew, we have to be willing to let people on, and maybe it’s implicit just by the fact that they’re deployed on a Verisk platform, but we’re not making any statements about the quality of a given model, the coverage of a given model. We’re not providing support for the models once they’re deployed. And very importantly, this is the critical part of the platform, is there’s absolute brick wall closure between the model IP and the platform and the model IP within Verisk. So we have no view of what they’re doing, and they have no view of what we’re doing. Provided they can work with our team to deploy on Oasis, that’s the value that we’re providing. We’re certainly not making any explicit statement about the quality of a model, and arguably not even implicitly, because I don’t believe we’ve ever turned anybody away. We’re happy to entertain a conversation.   

Matthew: Yeah. Well, it’s good to have that clarity. I can see how that makes a lot of sense. And then in addition to natural catastrophe models, you’ve got cyber now as well.   

Rob: Yeah, we’re really excited about that. So a few years ago, at one point Verisk did have a probabilistic cyber model, but we very intentionally took a step back and decided to sunset that because at the [00:09:00] time our clients were asking us to invest our limited resources in advancing natural catastrophe modelling. So severe convective storm was blowing up, for lack of a better word, around the US and clients were really struggling to understand that loss. So we reprioritised and stepped away from cyber. However, the Model Exchange is a great opportunity to provide the same sort of framework to deploy any model, whether it’s natural catastrophe or not. So we have signed partnership agreements with CyberCube, with Cyberwrite, with a company called Concinnity Risks, and we’ve been in discussions for quite some time. We’re working with Cyence from Guidewire to get them deployed also. It’s just a matter of getting the contract signed. So from our perspective that it’s exciting because you’re not independently exposed to one risk or another. Either you’re exposed or you’re not. So you likely have catastrophe risk from a property perspective, and you likely also have some degree of cyber risk. So Model Exchange will give you one platform where you can get a holistic view of all the risks to which you’re exposed in eventually one model run. So we think it’s pretty exciting. 

Matthew: And just a related question to that. Back when you and I were [00:10:00] sitting on opposite sides of the fence, for want of a better word, it was quite common for larger companies to have multi models. There was budget constraints and to some extent some convergence around some models, the more developed areas. Well, what’s happening these days? Are you still seeing quite a lot of multi-model licensing or people turning to go for one vendor and then adjust the models to reflect their own view of risk? 

Rob: I think what you just said is more common. My observation is increasingly people are centring on a platform, and because of the wealth of experience they have and their usage of models over years, they understand their own view of risk. And the models have evolved to a point where you can adjust them. There’s the opportunity to put your own view on frequency and vulnerability to get what is ultimately, and this has become the industry colloquialism of view of risk, so everyone can adapt the model to have their own view of risk in the market. There are a few cases. The alternate view, I would say, is that it’s rare for someone to have a full portfolio of models from any provider side by side, but you may pick and choose for a given region. So maybe you’ll have a Verisk model for US tropical cyclone and another model [00:11:00] for Australian tropical cyclone, for example. I’d say it’s more common to have one global suite that you adjust to be your own view. 

Matthew: Yeah, it’s good to hear that. And then for anybody listening that is thinking, “Oh, I quite fancy building a catastrophe model,” or they might already have one, we know quite a lot come out of research projects. What would be your suggestion as there is a focus now, the gaps where models don’t exist, but your clients are still asking for them, or as we think more broadly, not as clients, but areas of the world that may not necessarily be impacted by insurance, but still have a need for better analytics? 

Rob: Back to the catastrophe bond market, maybe global resilience. The Asian Development Bank used the Global Earthquake Model for a few under-modelled countries to get a catastrophe bond placement earlier this year. That’s a good example where that was successful. We talked about flood before. I think flood is a risk all over the world, and flood modelling is extremely expensive to create and extremely expensive to have truly be refined to a location level to help to get really granular and converged model loss analytics. So while we have a [00:12:00] very large flood team, the parts of the world where we model flood are limited, so that’s a good example where Model Exchange, companies like Fathom who are deployed on Model Exchange, would be a good alternative for people to perhaps deploy there. Those are top of mind. We continue to innovate, obviously, internally and add views of risk. We do not currently have, for example, a Canadian wildfire model. We are now building one of those, and we’ll get that to market in late 2027, early 2028. But it certainly informs our overall roadmap, knowing we have now Model Exchange on hand to see where does it make sense for us to continue to invest and build various views versus maybe it’s better to partner with someone on Model Exchange and sit side by side to have that view of risk be where we choose to go instead. 

Matthew: Let’s come back now and talk about the resilience and the protection gaps. So I know that in the past Verisk has given acknowledgement in the models or credits in the models for resilient measures that the ultimate risk holder, the company or homeowner, had in their property. When you talk about global resilience and [00:13:00] using the tools to help global resilience and risk mitigation, is that what you’re referring to in there or is there something broader than that? 

Rob: From our perspective, it’s bringing some sort of insurance or reinsurance product or some government backstop to provide protection or capital into regions where disasters may happen. Understanding, of course, that it’s ultimately an economic discussion, so models are a component of that. But we understand you need to have risk-takers, you need to have the underlying capital to supply the resources when the event happens. It gets tricky. We talked about parametric before. If you want to deploy a parametric instrument, you need to have some independent reporting agency to provide the hazard metric that can trigger the parametric bond or parametric insurance product, and sometimes those are hard to find, and sometimes they’re not established or they don’t have a history that would allow you to build an insurance product around it. You want to have validation data and see that it’s been around through past events and it can withstand the pressures of a very difficult hazard coming upon it. But we have one part in this equation, Matthew, which is [00:14:00] to provide some sort of analytic that allows a pricing mechanism to happen, some sort of assessment of what that risk is to bring parties together to have a conversation. Well, hopefully Model Exchange will help make that easier given some of the gaps where it addresses where Verisk doesn’t have models. 

Matthew: And so on that parametric topic, where do you see the areas of development on that? Over the last 10 years since we started InsTech, there’s a lot of companies started up doing niche products for parametric. Quite a few have struggled. Sometimes the reasons you mentioned, the indices aren’t there or they’re too expensive or they’re not trusted. But what are you seeing today? How has that space evolved in the last 10 years on the pure parametric side, which is itself a subclass, of course, of that whole insurance-linked securities area? 

Rob: It’s a great question. It’s the thing I’m probably asked the most often and have the most unsatisfying answer for, which I don’t see much evolution. If people have an opportunity to provide a product that’s not parametric, they tend to do that instead. So I’m not seeing it evolve to the point where I think maybe it was once promised [00:15:00] to. We see requests here and there, but they’re small, they’re bespoke, they’re very targeted, and it hasn’t come to a point where it is a true large-scale component of risk transfer, the way I think it was once envisaged to be. It just hasn’t materialised that way. 

Matthew: Yeah. A great idea, but for lots of reasons, really hard to get right. I say this as you say in the England or Britain touch wood, because I’ll say this and then there’ll be some big catastrophe in the next two weeks. Big catastrophes are actually quite rare, and if we look at those, and I think that’s part of the challenge for people building parametric products. They can spend money buying them, but then they’re not used, or of course, as we know, the basis risk is still quite tricky, so your protection doesn’t always pay out. 

Rob: I think that’s the hardest part. The basis risk is the hardest part. If you draw a box or a circle or you have some sort of a mathematical trigger, and you miss by just a little bit and you get nothing, that’s hard for people to rationalise, what you really need is some sort of capital in those events. I think that’s part of the reason why it’s been hard. Basis risk is just massive in these things. And 

Matthew: It does lead on to another topic we wanted to cover, which is the emerging markets and [00:16:00] protection gap because that’s where we are seeing more parametric solutions used, I’d say, because you just can’t get traditional insurance, and governments are starting to get a bit more comfortable using a parametric solution. But we’ve been talking about this protection gap for a long time. Are we actually making meaningful progress on filling in that protection gap or providing protection where there wasn’t protection before?   

Rob: Meaningful progress, I would say no. It doesn’t feel like meaningful progress. Every year we put out a paper on global model catastrophe losses versus actual losses. It has largely the same conclusion that the protection gap is surprisingly large in the United States because of the lack of earthquake coverage. It’s very large in South America. It’s very large in parts of Asia Pac, and we don’t really see it meaningfully changing for all the reasons we talked about before. So for our part, we’ll continue to shine a light on that. And understand, I’m not meaning to suggest at all that it’s easy and no one can just wave a magic wand and make this gap go away because it requires a lot. It [00:17:00] requires tools. It requires people to come to the table. It requires taking risk that maybe is not as well understood as it is in other parts of the world. That’s all fair, but as a risk modelling company, it’s our job to continue to illuminate areas where maybe there is risk that’s not currently protected and provide tools as best we can to help quantify that risk with the understanding that it’s much more uncertain for Malaysian flood than perhaps US tropical cyclone, for example.   

Matthew: Yeah. And I suppose part of it just reflects you and I having been around for a while — we’re always a bit more cautious about saying things are changing dramatically or the world’s about to change. But having said that, Oasis and now what you’re doing with the Model Exchange platform, and you alluded to this a few times earlier, that is making the models available to a broader audience. The Global Earthquake Model, GEM, you referenced. Feels like it’s still a big challenge to find the solutions, but actually the analytics and the tools are starting to be made more available, and tools that insurers relied upon are now gonna be available to more than just the insurance community. 

Rob: The tools are there, absolutely. And that’s a really valuable point. So I think the benefit of Model Exchange and where [00:18:00] we’re going with our core modelling platform, which is called Verisk Synergy Studio, will open up the opportunity for these analytics to go to places where maybe it’s been more difficult to get them in the past. So as a veteran in the space, you’ll know that historic catastrophe modelling platforms have been pretty large Microsoft SQL Server-based applications requiring large teams of individuals to operate them, a lot of data moving around. But a SaaS platform, a thinly deployed SaaS platform takes away a lot of that infrastructure requirement and opens up analytical accessibility to people who maybe didn’t have the teams to analyse it before. So governments, corporations, people can now have access to views of risk and tools that maybe weren’t in their toolkit before, and I think you’ll see us being pretty vocal about that over the next future horizon because we do believe we have the tools, and we now have the capability to get it into people’s hands where maybe it wasn’t before. 

Matthew: And talk about tools, Rob. Again, just playing back many years of experience and so we’ve seen a few things. What is your ground, based-in-reality view of a generative [00:20:00] AI, agentic AI application that you feel is really going to make a difference in the next 12 months, either within Verisk or if you just want to choose something from the world around you? Because we all see lots of enthusiasm, but when you’ve been around for a while and looked at analytics, you know that actually it’s quite a high threshold to be able to use any kind of analytics. So I’m just interested what your own personal views on what you really think might actually break through the noise on that. 

Rob: I will admit to you, I was probably sceptical for longer than I should have been, but I’ve seen our teams now deploying it across catastrophe and risk solutions specifically and Verisk more broadly. And the things that you can do with Claude Code and with AI solutions are dramatic. So I believe that within the next 12 months as we roll out Verisk Synergy Studio, we’ll be able to put into the market tools that actually provide natural language processing and, for lack of a better word, robotic interaction for very repetitive modelling workflows that let you engage with the data and engage with models in a way that just hasn’t [00:21:00] been possible before. And it’s not a dream. I think it is actually reality. It’s moving at a pace sufficiently fast where we can make this happen within a 12-month timeframe. 

Matthew: I guess what’s gonna happen, someone’s probably doing it already, but is you get agents can actually run the models as opposed to an analyst running the model, and then you can run the models, come back with different scenarios, different variations on parameters, and then process all of that in a much higher volume than you could have done before. It’s a bit of a brute force answer to my own question, but that seems to be one area I’d think was gonna make a difference, just in terms of the sheer volume that can be modelled. 

Rob: Client X’s agent calls Verisk’s agent and says, “Hey, I’ve got my data. Can you run my portfolio? Hey, what happens if I change my deductibles to Y? What if I eliminate these policies in this region? What’s my optimal portfolio look like?” And the agents can communicate with each other in a series of sensitivity analyses without someone having to wait for the machine to process or push data back and forth. That’s an envisaged agentic workflow that I think is very real. Imagine [00:22:00] hurricanes in the water where the National Hurricane Center puts out a cone, and the agent sees that it’s happening and says, “Hey, I’ve got this alert from the National Hurricane Center. Tell me what Verisk is saying about this.” And they can reach out to Verisk and get a pretty comprehensive view of radar data loss, footprint data, actual loss data for what’s happened so far, et cetera, et cetera. So it provides access to the intellectual property of a catastrophe model in a way that’s just much easier than it’s ever been able to be accessed before. And that will, I think, open up the door to maybe closing this protection gap in a meaningful way because it doesn’t require that really specialised knowledge that’s been built over 35 or 40 years of how to run the models, because AI can help you do it much more efficiently. 

Matthew: That hurricane sensing or hurricane sensor just gets you thinking you could have an analyst who wakes up in the morning and the model’s already run a landfall forecast because it’s picked up an early warning of a hurricane, and it’s run the model and says, “This is happening. And by the way,” rather than you having to scramble in the office, “this is your first estimate of landfall” — that whole thing gets really intriguing [00:23:00]. Talking about generative AI, I have now run my proposed questions through Claude to see what it thinks, and it came back with one that I said to you before, Rob, before I told you where it came from. I’m just gonna read it to you ’cause it was nothing to do with me. But Claude asked the question, so we’ll give Claude his chance to be a guest host on this podcast. And it said, “What is the honest answer to the skeptic who says that this Model Exchange platform is really just Verisk protecting market position as its own models face more competition?” Do you want to answer Claude on that? 

Rob: I would tell Claude that competition is a good thing. Competition pushes us to invest, to innovate, to think differently about these problems that the market has been attacking for 30-some-odd years. I guess I would ask Claude, “What would you have us do?” We have two choices. We can become more competitive and close ourselves more off, make it more difficult to understand what we’re doing, or we can embrace the competition with a true aim of bringing global resilience to individuals, communities and businesses, and make a mechanism where competition is [00:24:00] leveraged and put into the hands of the ultimate consumer. So our job is to provide risk analytics, and if we have an opportunity to provide more of those to more people to make the world better, my personal opinion, as long as they let me continue to sit in this chair, is that we’re gonna keep on doing that and keep on doing whatever we can to put more models into the hands of our clients. So take that, Claude. 

Matthew: It’s a great answer, and I don’t think we’re quite there yet, but I reckon in 12 months’ time, a little bit like our analogy with hurricane forecasting, you should be getting an email tomorrow in your inbox from Claude or some other generative AI tools listening to the podcast once it’s been released, so maybe not tomorrow. But once it’s been released, giving an answer to that ’cause it should be listening out there and come back and telling you automatically without us having to prompt it. I don’t think it’s gonna quite happen yet, but we can still ask it the question. And then a couple of ones, just one personal one though, Rob. So yeah, always great to hear people’s reading or listening preferences.  

Rob: Actually not much of a podcast guy, believe it or not, Matthew. I tend to read more than I listen to podcasts on the train. So I’ll give you a couple books that you’ve probably heard of, if that’s an acceptable response instead. [00:25:00]  

Matthew: Absolutely, like a bit of reading. 

Rob: One of my favourite books is Outliers by Malcolm Gladwell. It provides useful perspective into different ways to think about what actually drives success in people. And to me, some of the key takeaways of that book are relentless focus. I very much believe in decisive decision-making, intentional actions and relentless focus on the things that are important. And his research in that book suggests that has not gone to waste. If you really want to be great at something, spend the time in it and block out the noise and really focus. The other book that I really enjoyed is a book called Smart Brevity, which is how do you, particularly in today’s culture where there’s so much coming at you and so much difficulty finding someone’s attention, how do you really get to your point quickly, and how do you cut out filler words and cut out things that distract from your ultimate mission? I go back to this book every time I write a speech, every time I think about conference remarks. Because if you don’t get to the point right away, you’re likely never to get there, and people won’t stick around and wait [00:26:00] for it to happen. So that was a great read. 

Matthew: Two great books. I don’t know Smart Brevity, but I certainly am a big supporter, if not in practice, certainly in the concept of keeping things short and clear and repeating them. 

Rob: I don’t think so. I’d ask you the same thing. I feel like we had a really good conversation around the important points, which is we as Verisk are investing to bring risk solutions into people’s hands, whether they’re our models or someone else’s models. We believe we have a responsibility to help quantify catastrophe risk globally, and Model Exchange gives us a way to do that. I’m glad you brought up AI and agentic AI because it is moving at a pace that I think is challenging to embrace and maybe making people nervous. But I want to give your listeners the confidence that we’re on it, and they can rely on us to bring them solutions that are meaningful and practical in a way that they can deploy them. I think we covered it all, but happy to dive deeper on any topic you want to hit on. 

Matthew: And everything you said, everything with Model Exchange, your clients, potential clients, what would you offer to somebody in five years’ time, a company or a head of cat modelling [00:27:00] who’d embraced Model Exchange and doubled down on their relationship with Verisk? So somebody listening goes, “Well, what do I care? Does it really matter?” What are you offering them that’s going to make a difference in their lives over a five-year time period? 

Rob: I think we’re offering them the certainty that they’re doing everything they can to quantify global catastrophe risk from every angle that’s available in the market. And I mentioned before, we’ve continued to add providers. Adding KatRisk was a notable addition to Model Exchange. Adding cyber risk is a notable addition to Model Exchange. There are a few other providers out there that aren’t deployed on Model Exchange, but we’ll continue to pursue them. And over the course of the next five years, my hope is that we can provide a truly holistic ecosystem that represents all of the catastrophe risk or all the extreme event risk from anywhere in the world. And people who double down and invest in us will get a partner who’s very transparent, very open, very collaborative as to how we can work together with the world to help understand this risk in a better way. 

Matthew: Rob, thank you very much. I really enjoyed reflecting a little bit on where we’ve been, where we’ve come to, [00:28:00] and where we are going. Thank you for all the support from Verisk and all the great work you personally are doing in the company. Thank you very much. 

Rob: Thanks, Matthew. Great conversation. Appreciate it.