.png)
In this episode of Making Risk Flow, host Juan de Castro is joined by Marcus Ryu, Co-founder of Guidewire, to explore his fascinating entrepreneurial journey and the transformative role of technology in the insurance industry.
Marcus reflects on his nearly two decades of experience building Guidewire into a company that revolutionised insurance. From his early career at McKinsey to founding a startup in a traditionally conservative industry, Marcus shares his thoughts on overcoming the challenges of entrepreneurship, the importance of building complementary founding teams, and navigating the high-stakes world of sales. Together, Juan and Marcus also discuss the evolution of insuretech and its growing opportunities, Marcus’ transition into venture capital, and the pivotal moments in Guidewire’s history, including a legal battle with Accenture that nearly derailed the company.
Listen to the full episode here
Juan de Castro: Hello, my name is Juan de Castro and you're listening to Making Risk Flow. Every episode, I sit down with my industry-leading guests to demystify digital risk flows, share practical knowledge, and help you use them to unlock scalability in commercial insurance. Marcus, thank you so much for joining me today. It's a real pleasure to have you on the podcast. I'm really excited about the next 45 minutes of our conversation. I think you bring both an intimate knowledge of the industry and across many different roles. So I'm really excited. But let's start with an introduction of yourself. Give us an overview of your history.
Marcus Ryu: Yes, Juan, likewise, very pleased to be here, honored to be asked to participate. I have a fairly simple career. After graduate school, I ended up at McKinsey briefly, where I believe you were as well. And then worked for a software company here in Silicon Valley called Ariba. I was very, very successful in its time, though it eventually ended up as part of SAP. And then in 2001, completely to my astonishment, decided that I was put on this earth to be a software entrepreneur and left with five other co-founders and started Guidewire Software. And then somehow 20 years passed on a very long, arduous, but gratifying journey. The company went public in 2012. And I stayed on as CEO for another seven and a half years. And in 2019, I transitioned to the chairman role at Guidewire. And a few years after that, I decided to make a major career transition and become a venture capitalist, evaluated a few firms, but ended up joining Battery, the partnership here in 2022, which was very consequential. I'm sure we'll speak a bit about that. But I'm still closely involved with Guidewire. I, this year, transitioned out of the chairman role as well, so that I could focus on my role here at Battery. But I'm still quite closely involved with the company. In fact, I have kind of a notional title of strategic advisor to the CEO. And naturally, I'll always be attached to it. But what the day job is to be part of the investment team here at Battery, which is a 40-year-old venture capital and private equity firm that's been through many business cycles. Battery led the Series C for Guidewire back in 2007, which was the basis of the connection.
Juan de Castro: And I think we will unpick all those phases throughout the episode. Let's start with Guidewire. So after your management consulting experience, you decided to start a company in insurance. So from what I know from you, you were not close to insurance back then. So you were almost like looking for what is the industry that needs a solution? So tell us a bit more about how you ended up starting Guidewire.
Marcus Ryu: Yes. Well, there are two parts to the answer in the way you framed the question, which was you decided to start a company and the company was focused on insurance. Why start a company at all? That's an improbable decision. There's a lot to say about that. And I'm actually trying to write something more substantive, but it's a reckless act to decide to start a company. The odds are not in your favour, the vast majority of startups fail. And I think everybody knows that even if they don't always act rationally in light of that information. And you have to start with a kind of audacity to say that there is something wrong with the world. Not only that, but you have a solution to that problem. And not only that, that nobody else can solve the problem as well as you can. So it's kind of a threefold audacity to say that you should be an entrepreneur in the first place. And usually the motivations are not intellectual. The motivations are usually emotional and kind of deep in one's character. And one common trait that I've encountered among entrepreneurs is they just can't work for a boss. They're just too prickly, difficult, opinionated to work for a boss.
Juan de Castro: What was your case, Marcus?
Marcus Ryu: And that was certainly true for me. I was always contrarian in my opinions, always wanted to find the non-obvious answer always chafed under being managed, even when I was completely ignorant of the subject. So anyway, that was the character motivation, you know, just a desire to create something and do it on my own terms and be part of creating a culture. I didn't have to be the guy in charge, but I needed not to have someone else in charge. I wanted to be part of a group that made those decisions. So there were these kind of eye-minded, almost ideological reasons to start a company that had nothing to do with the domain. So that was half of it. The other half is what problem are we solving? And to put yourself back in the year 2000, 2001, the dot-com boom kind of reached its explosive peak, and then it collapsed. And the sentiment at that time was, at least among a certain set of entrepreneurs like us, was that we wanted to do something sober, something grounded, something substantive, where we were not trying to invent a radical new product or create a new industry out of whole cloth, but address an existing problem that was very challenging, that had been resistant to change, where people were unhappy with the incumbents. And we'd create something of really durable value. As you may recall, during the dot-com era, there was a lot of frivolous nonsense. It looks kind of like now at times. And especially post 9/11, there was a spirit of, let's do something really grounded, substantive, and maybe unglamorous, but ultimately durable and meaningful. So that was kind of a part of it. Another part was, we had a theory that vertical applications had been neglected, that there was a theory among venture capitalists at the time, which was you have to create a whole new software category that no one has ever imagined before. And you have to be able to sell it to everybody. And there were exemplars like the big ERP companies or Acebo that would suggest that that was the way to create a huge venture growth success story. And we had the opposite belief. It was that if you're going to solve a mission-critical problem, and something that's really vital to an organization, then very likely it's going to be an industry-specific solution. The things that insurance companies care most about tend to be specific to insurance. They're different, than what an oil and gas or a consumer packaged goods industry or a bank care about. They're specific to insurance. So our belief was that we should create a vertical application because we wanted to do something mission-critical, something that would be extremely durable and that we would have an advantage over much bigger companies like an SAP or Microsoft, because they didn't have the patience to really understand the requirements of one industry and build core applications for it. And that was, believe it or not, a contrarian theory in the year 2000, 2001. Now, I don't know if it's common sense, but partly because of Guidewire and some other exceptional successes like a Procore or a Viva, it's understood you can actually build a large venture-scale company just serving one industry. But I would take some pride in that we were a bit earlier in that. Now, why insurance? Well, part of it, there are these random things that happen in your life that end up steering the course of the decisions that you make. Then you look back and say, how arbitrary that that made all the difference. In my case, it was because my office mate at McKinsey in my very first year as a completely green associate was a former actuary at a completely random. It was like being sign, someone in your freshman year in college and over lunch, we would talk about insurance. And I didn't know the difference between property and casualty, and life insurance. I didn't know what is reserve was or what an underwriter did, and he gave me a basic education in these things. And if there's an interesting thing at that time, was that when I was at McKinsey, there was one category of project that was very lucrative for the firm. But every associate, every analyst, every junior member of the team absolutely detested. And that was an insurance closed file review. What that meant is that you would be deployed, perhaps you know about this Juan, where you would just be deployed into the bowels of an insurance claims operation, and you would be buried in paper files. And it was your job to analyze where mistakes have been made, and then make a case that this or that operational improvement would reduce the amount of indemnity leakage. And these were very lucrative projects for the firm and the insurance practice at McKinsey at the time, because it was kind of showing them the money. It wasn't like a theoretical process change. It was, you have direct economic deadweight loss that's coming from your claims leakage, and we will help you find it. And that planted a seed that after a few years in Silicon Valley, I thought, I proposed to my co-founders, I said, I think there's a problem here for one industry, that we can not only build an application that's more ergonomic, that's more flexible, etc., but that can deliver an ROI, a value proposition that's very tangible, and that's connected to a well-established theory of where there's an economic loss. And we can build a system whose proposition will essentially be, elevate your weakest claims or your most junior claims adjuster, and make decisions as well as the best claims adjuster, and thereby avoid claims leakage, and thereby make a massive difference to your operating income. That was the initial theory. And you can see it was like a combination of psychological factors, emotional factors, random bits of information that informed this belief. And we made many, many mistakes along the way, of course, and we can talk about some of those. But we were basically right about that. We were right about those issues.
Juan de Castro: So basically, you learned about the industry, understood there were painful processes with economic loss, opportunity to create something that addressed those pain points with an ROI, a clear ROI attached. But then, for somebody outside of the industry, you started creating a claim that you still almost knew nothing about claims.
Marcus Ryu: Indeed, yeah. I mean, we knew we knew little, we actually underestimated just how little we knew. One of the beautiful things about the insurance industry, of course, this is very large, and it employs millions of people, and you can talk to them. And if you approach them respectfully, deferentially, and ask them for their time, they will give it freely, as you express curiosity about what they do, and you ask their advice, and you iterate on. And that's what we did. We spoke to every person that we could network our way into in the insurance industry. And I don't mean CEOs. I mean, you know, a field adjuster, a nurse case manager, a clerical worker. I mean, they were not hard to find. Sometimes I see entrepreneurs, and they don't have the patience to do this. But this is vital. We really developed a kind of grounds-up understanding of not just the insurance industry, it was an insurance work. Like, what is the work that people do in the four walls of an insurance carrier? And we evolved that to kind of asking permission to do site visits. And we traded a little bit on our consulting backgrounds to say, hey, the reciprocity here is you'll let us shadow some of your team and make some observations, and we will deliver for you a report that will have those observations, improvement opportunities, et cetera. And it's all for free. We can't pay you. You don't pay us. But that was a deliverable. And so that helped us develop a real understanding of the domain. We also conveyed into that a belief that we're going to build a core system here. We're not going to build a little adjunct helper system that would help you make, I don't know, subrogation decisions or litigation management decisions. We wanted to build a serious piece of software. And that quickly led us to say, well, we have to be the transactional system of record for claims. We have to displace the claim system. And if we're going to build a claim system, then in the fullness of time, we have to build the policy system as well. That is the mothership that governs an insurance company. The whole assets of the insurance company are in a way instantiated in that policy system. And if we're doing that, then we should always also do billing while we're at it, because that's the other transactional system. And some insurers, all of those are just one big monolithic COBOL mainframe. But above a certain size these days, generally, those are now their own discrete application domains. But I would say very early in the company's history, within the first year, we had a theory, we were going to build all the core systems that define a P&C insurer. And equally important to concluding that was to say, we're not going to be anything else. We're not going to serve life insurers. We're not going to build digital solutions and mobile solutions. We identified the frame, the boundary of what our products would be. And essentially, that sustained the company to this day, 24 years later.
Juan de Castro: So you obviously thought you need to build the claims, policy admin and billing solution. Was that just different stages of your product roadmap? Or did you actually start with a version of the three from the beginning?
Marcus Ryu: No, it had to be claims first. We had to demonstrate. And there was an argument to be made. These sound silly now. But at the time, we had to make an argument, first of all, that the claim system is different from the policy system. As you know, on the original claim systems were just subsystems of a policy system. It was an attribute of a policy that had a claim on it. But we said, no, no, claims is its own system domain. And we're going to build a claim system first. And then secondly, it was, believe it or not, controversial to say you can build one claim system that can handle every kind of claim, handle personal auto claims, you can handle workers comp claims, you can handle environmental mass tort claims, all you need is one claim system. But this claim system will do justice to the variation to the heterogeneity across these claims types. And that was part of our value offering. I mean, these things sound very basic today, but they were big assertions to make in the market at the time. So we said we wanted to become a leader in claims first. And it was about four years in, excuse me, about four or five years in that we then embarked on the policy system journey. And that, of course, was a bigger mountain than the first.
Juan de Castro: I've heard you refer to your co-founders as your brothers, and also about the importance of having a complement, like a founding team or the leadership team with complementary skill sets. So I guess two questions on that. One is, how did you find them? So was it just like a group of people that you were looking to build a startup, or was it more surgical and thinking, okay, like we need somebody with a technical expertise? So how did the team get together?
Marcus Ryu: I wish that there was a more interesting Ocean's Eleven-like story about how that happened. It actually, it was just good luck. My co-founders came from two companies, Ariba, where I was, and then another company called Kana. And the bridge between the two was just a mutual friend who got us together. Interestingly, almost everyone on the team, we were all there kind of by product of kind of acquisition. Two came in from companies that Ariba had acquired, and likewise for Kana. But we identified in each other kind of a very, very strong intellectual affinity. But more important than that even was a kind of cultural or values affinity. We all were motivated to build something substantial. We were all restless. We were all impatient with a kind of frivolity, shallowness, short-term materialism that we felt, somewhat righteously, had defined that era. There's been a lot of silliness. A lot of capital has been spent on stupid ideas. There's been a lot of impatient, overexcitement, and hype. Let's do something that's the exact opposite of that. Let's do something grounded, substantial, long-term, radically transparent, et cetera. And the fact that we were also well, aligned on that culturally made a big difference. But as you point out, and I've talked about this before, it was not enough just to kind of have the same value system. We actually had to have complementary roles. And there I can only attribute providence or good luck. I had John, who started the company as CEO, was absolutely brilliant as an early sales leader. Ken was a genius as a product manager. He defined what our products would do based on all of these meetings and visits that I described. Another John, John Seibel, was our CTO. And he made. He made exactly the right kind of architectural decisions that it took to build the product with the state of the art at the time. Mark, who also came from Kana, was a brilliant first engineer and eventually led most of the engineering team in building the platform. And James, who also came from McKinsey, like I did, was fantastic at marketing with almost no assets. So the question is, what did I do? What was my role? My role was kind of funny. I take some credit for the idea. It was my idea based on, as I've already told the story here. But I could not build software. I was not trained as a software engineer. I had no experience selling anything in my life. And you can't just spend all your time strategizing. You have to execute. And so I quickly realized if I'm going to be of any use here at all, I have to learn how to sell. I have to learn how to get into the market and persuade people. And that was very, very humbling for me. Very difficult. Because that was not my nature. I was very cerebral, very well educated. I was, you know, I made a reasonably good fit as a consultant, but I was kind of useless as a salesman. And Guidewire gave me many, many things. The whole journey was profoundly meaningful. But sometimes I think one of the most meaningful things it gave me was it forced me to learn how to relate to people who are very different than I am, who learned to find common ground with them, learn to meet their needs and learn to become a salesperson. And I approached sales with a kind of disdain, a kind of completely elitist snobbery. And now I regard sales as a noble profession and I'm proud to be a salesman. I think I'm probably better at that than most other things. And I learned that by necessity in the Guidewire journey.
Juan de Castro: And at what stage did you move from founder-led sales motion? At what point did you start by building a professional team of salespeople?
Marcus Ryu: Well, I'll tell you, I left at year 19 and we were still founder-led sales. Because the nature of what we sold was so mission critical. It was so high stakes. Any executive that sponsored a Guidewire project or a core system transformation was betting his or her career on it. And the project would be very expensive, take a lot of capital expense, it'd take a lot of talented people's time and divert them into doing a huge transformation program. And the track record of these programs, not Guidewire projects, but projects in general, was terrible. Almost every company we talked to had scar tissue from two, three, four failed projects to try to rewrite their policy system or transform their claim system, etc. So it was very high stakes. And if it's going to be high stakes, that means you had human beings that were taking risks with their career. And if that were the case, to persuade them, you cannot just say, hey, we're here for a transaction. We've got a good product. You had to commit something more profound. You had to commit a readiness to go whatever distance it took to make this project successful. And that is something that a salesperson can't really say, even if a very talented salesperson can't really do that. It has to be the principles in the company. It has to be often the founders and the CEO. And so even deep into the company's history, even after we had a professional sales force, I found that it was extremely important for me to kind of commit the company, commit myself, to the success of projects, to do whatever it would take. And sometimes it was just enough to say that and things went well. But sometimes it wasn't. Sometimes things would go wrong. And it would require following through to do whatever it takes to make those projects successful.
Juan de Castro: One of the things I've heard you say in the past, which resonates quite a lot with me, is like this concept of the false summits. I think you call them, right? Which is you're always thinking, well, once we get the 10th customer, this will just be a factory, right? Or like of growth. Or once you, whatever, expand into Asia, this will just run by itself. Can you talk a bit more about that?
Marcus Ryu: Yeah, sure. So, I mean, it's so difficult when you're starting a company, when you have no assets and you just have a dream and everything seems impossible. It's like, I remember thinking, when we ever get a customer to pay for us, pay for this software, that felt like an impossible dream. It was so hard just to get people to spend, just to meet us, let alone spend time with us, let alone commit the time of their team to do something, let alone commit to doing some kind of even free project. All of that felt so hard. How will we get them to ever evaluate and then buy our product? I mean, it just felt like impossible. And then how will we get 10 companies to do it? And all of that felt so terribly impossible that we just had to, as a matter of psychological survival, had to just kind of push that outside the frame and say, the goal is not to become a billion-dollar company. The goal is not even to sell anything. The goal is to build 10 good relationships. You have to redefine the goalposts in a way that would allow you to make progress and not be completely demoralized. And one part of the fiction of that, necessary fiction, was to say, well, but once we achieve that, then we can have a whole new horizon of action. We'll have all these new things that are so difficult today, will be much easier later on. And it would be after the first customer, things will be easier. After the first three customers, after the first customers in live production, then surely. After we've demonstrated we can support any line of business. After we've shown that someone will buy claims and the policy system, because that would really prove that they had great confidence in us. After we've shown that we can handle a tier one insurer and not just a small regional insurer. Infinite series of these. And every one of them did not unlock some. The work never got easier. It never, ever got easier. But what did get easier was the specific task in question. Actually, there were days in the beginning when there were questions about our viability. Will you even exist as a company next year? And there was a point when that question was no longer asked. Does your claim system actually work? Can we bring this? And there was a point when that was not asked. We were in another trench fighting a different battle, struggling. And so the work did not feel any easier, but the truth was things felt so impossible actually had been conquered. And what I learned was it was important as a group to kind of, in the midst of the struggle, to turn back and look at the things that had seemed so impossible before, but were now in fact conquered. That we're now part of our assets, part of our achievement in the trophy room. And so we have to stop and look at that because otherwise we will feel that we are digging a trench to infinity and there'll never be a moment of gratification. That's one lesson. The other lesson is you have to find joy in digging. If you were focused on a destination, then that is a pure formula for misery on the entrepreneurial journey. And now I'm a venture capitalist. I talk to entrepreneurs all day long. And the ones that I worry about sometimes are, they have a conviction of a destination that is imminent. They say, as soon as this happens, then we'll be here. We'll be doing something amazing. And I much prefer to hear than take joy in the domain. They say like this problem is so important to solve and look at the progress we're making and solving this problem and how much value we can create as opposed to how fantastic their company is going to be, how valuable they're going to be, how much revenue they're going to be making. It's a subtle difference. But, I tend to find that it matches a kind of character difference that is not really right or wrong, not good or bad, but is connected in my opinion, to the ability to persevere through the difficult task.
Juan de Castro: And I guess after you were there for 19 years, is that right?
Marcus Ryu: That's right. And then we've been involved still for the last four or five years.
Juan de Castro: Yeah. But I guess even when you had an executive role, even after 19 years, you didn't feel anything. We were saying you didn't feel anything. It was just a different set of challenges and problems to solve, right?
Marcus Ryu: It certainly never felt easier. In some ways, it felt harder for different reasons, but it never felt easier. And God, where today is over a billion, 1.1 billion or so in revenue, this number would have seemed to me like science fiction in the early days. Believe it or not, we said that God, will never go public. It's not that kind of opportunity. You know, we try to keep our expectations so low. It was so important to keep them low so that we could not be disappointed, so that we could just focus on the problem. But it never got easier. But there were exceptional rewards, economic, yes, but exceptional rewards and satisfaction in having solved hard problems that people did not believe were solvable through that journey. And even today, I think the company has actually more opportunity than even the most optimistic person about the company was me, I could have believed in. And that's humbling in its own way.
Juan de Castro: And if you thought you would never go public, then you IPO, a very successful IPO. What drove that decision?
Marcus Ryu: As I sometimes say, you shake the hand of the devil, you're investors, and then they own the company, not you. That was certainly true for us. I think founders today maybe have a lot more leverage than we did at the time. But we gave away half the company in our Series A for $4 million. People don't, I didn't give it away. We sold it, I guess you could say. When I talk to entrepreneurs now, they can't believe it. It seems absurd, that you would do that, that six guys would do that and then continue to work for another decade and more. The terms of the game have changed. You could say it's good or bad. I have my own opinions. But there was no question that once the company was now succeeding, it had reached $10 million, $50 million, $80 million, $100 million in revenue, that we were going to go public. I talk sometimes about how much more difficult it was for entrepreneurs 20 years ago, 15 years ago. There was one respect in which it was somewhat easier. And that was that the bar to go public was substantially lower than it is today. For whatever reason, whatever market dynamics and structural reasons, the threshold for IPO is extremely high right now, almost prohibitively high. And that may change or it may not. But we did meet the criteria about a decade in. And even then there were questions. The global financial crisis had happened. There were not a lot of IPOs. We were, I'm proud to say, kind of considered at the time to be one of the IPO that kind of reopened the market in 2012. We went public at a valuation of about a billion dollars. And to me, that felt like a fantasy number, at the time that that was possible.
Juan de Castro: What was the ARR at the time? I'm not sure if this is public or not, but…
Marcus Ryu: We had services revenue as well. I think our ARR was in the 90 million-ish range. It's all public information. I don't follow exactly. The revenue of the company was around 120 or 130. It was approximately 100 million in ARR.
Juan de Castro: One last question on this first chapter on Guidewire, which is fascinating. And obviously, I'm also personally very interested in it. But the last thing I've heard you talk about in this Guidewire experience is how painful events led to a better outcome. I think one of the examples you mentioned was like a, I think it was your first customer churning. Can you give a couple of those examples? I thought those were fascinating.
Marcus Ryu: There were a whole sequence of things that happened over the history of the company that each of them felt like a calamity. Like this is the worst possible thing that could have happened. And yet, miraculously, sometimes turn out to be the best. There are a couple that I've talked about in public before, and I'll share one or two others that our first customer was an outgrowth of one of these consulting projects that I described. And we got them kind of agree to this sort of provisional version of a sort of license implementation of our product. And we were thrilled and they were going to pay us with tons of milestones and contingencies and so forth. But something that vaguely resembled a software contract. And we had a project and I was on the team and we were there day in and day out trying to implement. And on the basis of that, we raised our Series A where we were thrilled to sell half the company for $4 million. And now we were financed. And about six months in, they just said, change our minds. We can't do this. The reasons aren't really important because they had nothing to do with us, but we were cancelled. And at the time, calamity, like what are we going to do now? Then amazingly, two things happened right after that. Within a few weeks, number one, we sold our second customer and that was extremely important that we didn't have the first one anymore because that one was so hard to implement. There was so much work to do that it actually required all the energies of the team, including myself to just live essentially in Chicago for almost two years to get that program to work. But the second thing that was important was that the scope of the project that we had sold of the implementation of the offering was wrong. It was wrong, not because we had wanted to, but kind of the way that negotiation and the situation was such that they wanted to implement our product as a front end to their existing claim system. And it had a certain logic to it, but it was flawed logic. We tried to explain that. But in the end, it was like either we capitulate on this point or I don't have a customer. And so we could get related.
Juan de Castro: So the difference was being a front end versus replacing the system?
Marcus Ryu: Versus replacing the actual core system. Essentially, there were other complications and nuances, but that was the essence of it. And it would have been terrible because we would have had the biggest problem was financials, because the financial and you had one system. Namely, ours that you could generate financial initiate financial transactions. But the existing system of record was going to be the financial source of truth. And financial transaction is not just a number. It's a state. It's a very the claim is in a certain state. It's in a state of being evaluated. It's closed. It's being reopened. And the system of record was stateful. It could kept track of the state, whereas our as a front end system was kind of a slave to that back end system. And that led to all kinds of incredibly difficult and problematic integration issues because you did not have one system. Our system was not in control of the state of the claim. And anyway, it was a flawed implementation model. Who knows? But it probably would have led to failure at some point, but not immediate failure. We would have gone here and plus and then fail. And we actually dodged a bullet. The worst thing that happened was that we thought we fell into a pit. But in fact, by falling in the pit, we avoided the bullets that was going to kill us. And that was only one of those moments. It happened in other fashion. The other one I've talked a lot about was we were sued, by our primary competitor. This is now about eight years into the journey. That was by Accenture. That's the same huge consulting firm we all know. At the time, they had a theory that they could use their patent portfolio as an offensive competitive weapon. And Guidewire was their test case. We were the first example of how they were going to apply this new IP first, competitive strategy. And we were completely shocked. They announced the lawsuit like two days before Christmas, and we thought it was a joke. At first, we didn't know what to do. And the next thing you know, we couldn't sell any software for a year, year and a half. And it was because our customers would say this lawsuit has no merit, but we can't buy it and then be prevented from using it. So solve it and then we'll be fine. But we couldn't solve it because Accenture had no interest in resolving the issue. They had used it as an aggressive competitive weapon. And again, what could possibly be worse than this? We had no defense against this attack. There was nothing we could offer. There was no way to settle it because we tried. But they said, that was not the goal. The whole purpose of the lawsuit was to just put us out of business. And we felt we have no cards to play. It is game over. And this was almost 10 years in. And I've talked about this in public, but it's very hard to describe this to people who haven't experienced it. When you've committed yourself and sacrificed everything you have for like a decade, and then you have nothing to show for it. It was such a dark time. There was a very complicated story about how we ended up defeating the lawsuit. And in essence, we had very creative, clever ways to figure out. Oh, we kind of have a counterattack and we mobilized our customers behind us. And we could spend another podcast just talking about that. But anyway, in the end, we won the lawsuit and we were able to achieve a settlement kind of on our terms. Everything was over. And that was somewhere between three and six months later, the company went public. And again, it turned out completely amazingly that that was the best thing that could have happened to us. Because even though we were so full of angst and frustration about trying to solve the lawsuit. The fact is that. Meanwhile, the company was maturing in other ways. We had evolved our product in certain ways. We evolved our implementation methodology in certain ways. We had built up more customer goodwill and we had done more in terms of IPO readiness. By the time we went public, we were actually a stronger, larger, more resilient company than we would have been having gone public two years, three years prior as our investors had wanted us to do. And in the 15 years since I've thought about it many times, the way that we dreaded and feared and thought was the worst possible thing that could happen turned out to be turned out to be helpful to us. And so I think the lesson cuts both ways. Something terrible turns out to be not so good. And similarly, there were moments of ecstatic celebration that were turned out to be completely premature that were wrong. We, I don't want to make sure my names, but ones where we thought, oh, we've finally arrived. We've landed in this new country. We've gotten this huge name to commit to us, but there was something wrong about the way that the project had been defined or whatever it might be. And it turned out to, to cut the wrong way. And so overall lesson from it, of course, is to just is to find some place in the middle, to find a kind of equilibrium and a serenity in the journey, as opposed to riding this intense emotional roller coaster that's part of being an entrepreneur that is always careening from ecstasy and to catastrophe, because it's too hard. It's too long a journey. And I think the best entrepreneurs that I've met and what I learned in my own journey was the importance of finding like a kind of inner equilibrium that does not overreact in either direction.
Juan de Castro: I think that is a fantastic way to wrap up this first act on Guidewire. And I really like your point about the importance of serenity and just like almost smoothening both the peaks and the valleys. So like, do not get overly excited, but also when things go wrong, just have resilience. And it must have been really tough for you, especially when you see a competitor going against you with no real reason, but there's nothing you can do about it. It has to be really heartbreaking.
Marcus Ryu: Yes. I mean, actually there's an insurance-like lesson in it. I'll tell you, like insurers, the nature of the business is that you were insuring against terrible events and you have a catastrophe that happens and you would think there's a disaster. There's tens of billions of dollars are being lost, but that is part of the model that is going to happen. And then you have a year when things look very serene and it seems like, you know, as an insurer, you're just printing money. Well, yeah, but inevitably there will be catastrophes, there will be losses. And the super successes in insurance, the Buffetts and that tier, have that kind of serenity. They say, look, what's important is that our checks cash when there's a disaster. There will be good times when we harvest and there will be times where we have to pay. But if our underwriting is disciplined, we don't get over-exuberant, we don't chase premium growth at any cost, we'll be here for the long-term and it'll all work out. And that's a very serene kind of point of view, which is not very Silicon Valley-like, I'll tell you.
Juan de Castro: Definitely. Okay, so then after 19 years, you move on from Guidewire. You mentioned you stayed as chairman for a while and now as an advisor. But then the first question is, okay, so I guess once you moved on from the CEO role, did you know what you wanted to do next?
Marcus Ryu: The default course was to start another company. The Silicon Valley dream is that you work as an employee at some good company or companies. Then you start your own company, but you're poor. You have to rely on other capital. And so maybe you succeed and that gives you a basis to then have your third act and start a great company, but now on your own terms. That's the kind of mythical career arc that a lot of Silicon Valley people want to live. And I had a similar aspiration. And there were two reasons that deterred me from that. Number one was just sheer fatigue. I was very tired after 19 years. I think if it were today, it would be different. But at the time, it wasn't just a question of video conferencing technology. It was not the social norm. So I had to be in person everywhere all the time. I flew at least 250,000 miles a year. And most of that was just in the U.S. or in North America to the Midwest, to the East Coast, to Canada, to the Southwest. And that took an enormous toll on me physically and emotionally. I paid a high price in my health. And it was just a long time to be stressed. And sometimes the relationships with our board were not ideal. And after two decades, it felt like it was a very high price. And I've never been more energized than I am now. But at the time, at the moment, I said, if I don't have an interval here, then I'm going to pay a price that maybe I cannot afford. And so I recalled that and said, I need to take that seriously in terms of my other commitments. The second reason is more positive, which is during the first decade or so of the company, before the IPO, we were just in a tunnel. And I had my brothers and people in the company. But I didn't have any sense of what other entrepreneurs were doing. I just, there was no time. And there was no real context for it then. You know, today, the world's very different. You have incubators and meetups and conferences and all these other settings where entrepreneurs meet and kind of console each other and help each other. That didn't really exist then. But after the IPO, I got involved with a host of other private companies, joined some boards. And I realized that there was a deep universality in the kind of experience that I had had and was still having, and that I found enormous gratification in interacting with other entrepreneurs. Not necessarily like they're better human beings, but they are a different kind of human beings. They are a subset of people who have extraordinary passion, sometimes unreasonable, sometimes not, of a kind of intensity that I just want to be around, that's my tribe of people. And I want to spend as much time around these people as possible. And maybe being an investor is a good way to do that. And I can be a little different than the average institutional investor, and that I'm not, yes, of course, trying to underwrite and find good deals, but can be a resource, can be a different kind of more empathetic, somewhat an investor, more capable of inhabiting that first person habitus that you have as an entrepreneur and to help them. Basically, the kind of investor I wish I had along the way. And it was happening somewhat at random that I would meet companies. But I said to do this properly, I need to be attached to a platform like a world class investing platform. And long story short, that led me to join Battery. Those are the main motivations. Now, the venture industry is so different, I mean, I thought I knew it from the outside. After all, we raised three rounds of venture capital and so on. It's radically different now than it was then. The biggest difference being it's just much larger. It's probably 10 times the size as when we raised our Series A and B in the early 2000s. And it's hyper-professionalized. So there were no analysts and associates, or at least not anywhere like there is today, that are so systematically combing the entire universe of startup companies. And so what that means is it's very efficient. It means that the valuations are, to my mind, astronomically higher than they were during my era and vastly more competitive to be a VC. So, you know, my joke is I was an entrepreneur when it was much better to be a venture capitalist, and I've chosen to be a venture capitalist when it seems to be much better to be an entrepreneur. That's okay. I have no regrets or complaints, but it does seem that way. Nonetheless, it is, I think, a glorious time to be an entrepreneur or a VC for that matter. It's a time of incredible change, of just incredible connectedness between people doing different things that have relevance to each other. And the kind of learning that I've been able to have in the last few years, just from one company, learning from entrepreneurs, learning from experts that are advising about that, has been fantastic. It's certainly much more competitive than I would have ever imagined, but that has its own rewards too.
Juan de Castro: And having been an entrepreneur yourself, and obviously having good experiences and bad experiences with investors, has that informed how you act as an investor now? And in what way?
Marcus Ryu: Very much. I mean, I want to be the investor that, as I said, that I wish I had. I think when you're an entrepreneur, right, the thing that's very irritating is to just be pinged continuously by other investment firms saying, tell us your numbers, essentially. Maybe they put it nicely, but like, tell us your numbers, tell us your numbers. And so that they can figure out, do they want to make a bet on you? And that's very irritating. So my point of interaction is always, let me internalize your theory of the world. What is your diagnosis? What are you contending should exist that doesn't exist? What is your strategic account here? And I really try to inhabit that from, again, like the first person perspective, and ask questions as deeply as possible about that strategic thesis, before jumping to understanding the financial case. You have to do a certain set of, financial forensics to be a good investor. And Battery does that exceedingly well. And I'm proud to be part of a firm that does it as well as Battery does. But my own distinctive approach is to complement that as early as possible with a deep appreciation and understanding of the strategy. And then to underwrite kind of more, to include, or let's say, make a larger part of the underwriting, the human dimension of the leadership team, their deep motivations, their resilience. Do they have the kind of character traits that, in my opinion, are dispositive of entrepreneurial success?
Juan de Castro: Because I guess the rest, almost like the numbers, it will flow from the quality of the vision, the strategy, the team, right?
Marcus Ryu: Yeah, and mindful that the Guidewire was not perfect on those dimensions. I mean, we had very lumpy sales. We had perpetual licenses. I mean, some of them in the early days, which were ugly even then. Even 20 years ago, those were ugly. And we knew we did not want them. We wanted term licenses, which are kind of a precursor to what subscription revenue is. We understood that very well. But we were ugly in that way. We seemed to spend too much on R&D and not enough on sales. And our investors were trying to get us to look more like an early Siebel or something. And we said that wouldn't make sense. We would be pushing on a string to hire more salespeople. The product hasn't achieved its, first of all, it's not mature enough yet. And our market position isn't mature enough yet. So just more selling is not the answer. And we had conflict over that. So in short, I try very hard not to apply a kind of simplistic SaaS template to what success looks like. And some of the most interesting companies solving the most fundamental problems, are not going to look like beautiful portraits of SaaS metrics all pointing in the right direction. There are companies that do look like that, but that's not the only kind of success.
Juan de Castro: So now I know that now at Battery, obviously you bring a lot of the InsurTech insurance expertise, even though you're not, I mean, you've said in the past, you don't want to be the insurance guy, but you're quite involved in investing in insurance. When you look at the next generation of providers of technology and insurance, first of all, how is it different from the previous generation from those companies 20 years ago?
Marcus Ryu: Cool. So there are vastly more InsurTechs than, well, the category didn't even exist as a term of art 20 years ago. That's for one, but there are vastly more insurtechs that are addressing, well, let's see, as you know, there are several categories. There are ones that are risk bearing or risk gathering, MGAs and the like. And then there are those that are selling technology or data to the industry. And then there are a few other very adventurous ones that are trying to create kind of multi-sided marketplaces. But broadly speaking, it's those first two categories. I think there are interesting opportunities in each of them. Really tech forward MGAs did not exist, at least not to my knowledge or awareness 15, 20 years ago. Of course, they were MGAs, but they looked very different. Now you have Silicon Valley style technology first entrepreneurs that have a theory of a different category of data, a different kind of underwriting signal that they can say, we'll find capacity and go to market with it. And the cyber is probably the purest example, but there are many others that are almost every category, almost every line of insurance has that. And in fact, the first investment that I sponsored here at Battery was in Fairmatic, which does smartphone based telematics for commercial auto insurance, which I think is a fantastic opportunity. So I think that notion that you could find a different kind of signal, you know, build an underwriting advantage and go to market directly with that is new and relatively new. And there are many, there's a proliferating set of those opportunities that some of which obviously I believe are venture investable, not all of them, but some of them are. And one interesting thing about those from an investor standpoint is they don't exactly look like your typical SaaS venture backed company. They lack some of those qualities, but they have some other fantastic advantages. One of which is there are many acquirers. If you are a hundred, $200 million DWP MGA, you know, that with good loss ratios, then you don't just have one or two potential strategic acquirers. You have, you know, you have dozens and that's a very positive thing from an investor perspective.
Juan de Castro: So would an example of that be like the Corvus and Travelers acquisition?
Marcus Ryu: That's a notable recent example. And I think there'll be many more. Where let's just say in a line like commercial auto, it's recognized that to underwrite this well and price it well today, you cannot do it just through conventional approaches. You need to bring a really like telematics first risk selection and pricing approach. And therefore an insurer that wants to, a conventional insurer, an incumbent insurer that wants to do that may very readily conclude that the best way to do this is to acquire an MGA that knows how to do it well. And I think that'll be true for many other lines of business as well. On the data and tool side, there's data. And of course, there are many new categories of data that are very interesting. You know, there's earth observation data, there's social media data, there's all kinds of cyber relevant data that are that are very insurance relevant. And I think some of these are horizontal, but I think they're ones that are really trying to do that in an industry specific way. And data businesses, I think they're well understood in terms of how they ought to be valued. And I think those are of interest. Personally, I find the greatest interest in that when the various new kinds of technology or applications. And within that, the most interesting area to me, I think is the most interesting opportunity in the moment is related to pricing and pricing and say advanced underwriting. So the second investment that I sponsored at Battery is in a company called Hyperexponential based in the UK, as you know. And they have an advanced pricing decision intelligence tool that they started with extraordinary success in the complex London market, complex commercial syndicate business. And the reason that this is such a great moment for that platform is that the role of the actuary is really evolving. It's not just experienced and retrospective based. But increasingly forward looking, building fundamental models that are trying to really model the exposures and using machine learning, working within Python rather than Excel and coming to a fundamentally first principles reappraisal of what risks look like. And that means that the actuary himself or herself is becoming more and more like a programmer, more like a developer and who needs an IDE to work within. And that insight is what Hyperexponential was extremely successful in bringing to market. Now, they have a lot of work ahead of them and not all lines of business are quite as receptive as London market syndicate business. But I think it's an extremely fertile area that will transform the industry.
Juan de Castro: And it's a fantastic team, too, which I know well.
Marcus Ryu: Fantastic team. And they are the kind of classic, unlike me and my co-founders, they came from the domain. And that's the other classic path to building a company is you develop the tool that you wish existed for yourself. And you're just so frustrated it doesn't exist that you have to go create it yourself. And Amrit is very much in that mould, and doing a great job. And then this is other extraordinary phenomenon that your company is right at the heart of, which is the digitization of risk. And there's been this extraordinary, magical advance in LLM that LLMs have brought that have broken down this impedance that language provides. You know, the language creates and has suddenly made, in many respects, natural language interpretable by machines. And this has very profound implications. There's so much impedance between us and the tools that we use all day. You have to express yourself in just the right way. And the data has to be in just the right format in order to fit into systems and to get them to do what they want. And there's this extraordinary potential now. And I think Cytora is right at the forefront of this in insurance where natural language can eliminate that impedance, where you can just describe what you want done. And the semantics of that are understood, sort of, so to speak, or synthetically understood by an LLM powered system. And then it knows the actions to take on the other side of it. And it handles all the ugly syntax and all the frustrating syntax it takes to get other systems to do what you want. And I don't think that's just in insurance, but surely insurance, among other industries, will be transformed in the near term by this advance. I think it's particularly relevant for insurance because you have so much unstructured data that creates an opacity through the value chain between the primary risk, between the broker, the primary carrier, the reinsurance broker, that classic value chain. And every step between them, it's very lossy. You have a kind of translation, retranslation. There is a lack of a universal standard. And it's a very, very hard coordination problem to create those universal standards. So it's not really realistic that they'll be created. What the kind of approach that Cytora is taking that I find so compelling is that you don't need to create a universal standard as such. You can create a translator between each of these stages that are mediated by AI. And you don't need to entrust AI to make fundamental insurance decisions. In fact, I don't know of any insurer that's really doing that to any serious degree. But you can leverage AI massively to reduce the impedance of getting data from one counterparty or from one system to another. And there's huge value creation opportunity in that.
Juan de Castro: And I think you're totally right. When you look at the inefficiencies of the value chain in insurance, a lot of it is about requiring multiple interpretations of the risk data that moves across the value chain. This is why we are so passionate about what we do, is the concept of being able to understand completely an unstructured description of a risk and translate it into the view of risk of each insurance company. And I think one of the reasons standards have not really picked up is one is, well, in a many-to-many relationship, it's quite difficult to create the standards. But I think even more importantly is every insurer wants to have their own view of risks. And that's what creates the competitive advantage, right? You want to look at risk differently from your competitor, which again goes back to your thesis about these new cyber MGAs, so that they look at risk differently. What you're saying is you need to enable insurance companies, but it's to look at this differently, but overcome the challenges of having to understand different definitions of risk.
Marcus Ryu: My term for it is impedance. Like you get a submission, a commercial policy application, and it's very intricate. It has lots of unstructured information. It has complex schedules, which are in non-standard formats. The columns aren't always in the same order, et cetera. And you have to make a judgment. But just to get that in a state that you can even judge it is very demanding, as you know. And so before, there was no shortcut. Because there was no universal standard to which everyone would map. It was a completely fanciful idea that you could create such a universal template. Because the superset system would be a data model that was completely unwieldy. You may know that there were been attempts. For example, IBM in the early or mid-2000s had created something called IIA, which was their attempt to build on Accord and other standards and build a kind of monumentally complex universal PNC insurance data model, which got precisely nowhere. Because it was completely unmanageable. And to try to build a system that would instantiate this theoretical model was... No one was willing to do it, and for good reason. And then the user experience that you would create out of trying to navigate this massive model, huge portions of which may not even be relevant to you, it was just designed to fail. So you can't do that. You cannot create a superset model. But you need somehow to translate this unwieldy corpus of data into some form that is first intelligible by you, but also reasonably normalized, so that it's not a separate, unique adventure every time you want to look at a new risk. And so that requires some kind of data standard. The problem is that the existing systems don't have that model either. And that includes Guidewire. Like the Guidewire policy system, you don't want to create all of the data elements. That it would take to fully evaluate every risk, because not all of those are transactionally relevant. So you need to have kind of a staging area in between where you can normalize that understanding of what the risk is to make decisions on it. But then do you really want to create a whole new database, a whole new application to do that? Some companies have concluded that they should. Those are very mixed success. But I really love some of the thinking that I know that Cytora shares is that, well, you should make it very easy to create this kind of staging area. You should be able to describe it in natural language. And then we should make it easy to translate this unwieldy, complex submission, for example, and get it into this staging area where it can be evaluated. I think this is a fascinating idea. I think that it's something that could not be done 10 years ago. It is just, that the prerequisites did not exist. And I think it's very unsurprising that insurers are really excited about it today.
Juan de Castro: So, yeah, I know you also often talk about avoiding solutions, looking for a problem. And I think that the example you just gave about what we do, you say it's a great example of a problem that required it, that didn't have a solution 10 years ago. And I thought, have you seen any other areas where large language models are being applied in insurance? Or is there any?
Marcus Ryu: I think there are lots of places where they can be applied. I mean, just interpretation of documents, which is not necessarily just for underwriting, but I think this gets overstated sometimes. But nonetheless, there's still a large volume of documents that get ingested by insurers all the time. And to have those not just parsed and digitized, but I mean, really digitized, I mean, transforming the structured information or to have the semantic intent extracted reliably from, I think there's a lot of value in that. There are certain kinds of decisions that, especially at the low, relatively low level, that I think the current state of the art can handle. So for example, every insurer has an underwriting manual and it expresses a set of intentions about the kind of risks that comprise its appetite. And that LLMs are certainly at the state where they can understand, so to speak, synthetically, quote unquote, understand that intention and then apply that intention to new information and then judge what is the likelihood of a fit and appetite. And you could call this, this is like the first step of AI driven, underwriting, but I think that's in a way kind of glorifies it beyond what's necessary. I think you're making a determination that is not terribly complicated, but is very tedious in the first order. And I think LLMs today are certainly at the state where they can do that. I think it's not just LLMs, but the general universe of natural language, NLP techniques to evaluate other very valuable, unstructured information in a claims operation. So claims notes, for example, are being mined very successfully on, on complex long tailed, very expensive claims, you know, that usually involves bodily injury or casualty of some sort, where figuring out the right, determining what is really going on here. Is this claim being well handled? Is there any prospect of this injured worker going back to work? Or what's the disposition that we should come to here? I think that NLP and LLMs are certainly at the level where they're finding really useful signal in that. And that's valuable because the kind of veteran claims adjuster, claims examiner, as they often call them, in workers' comp that can parse all of that, read it, come to a good judgment about it, and do that efficiently over hundreds or thousands of claims. That's a very scarce skill. And so there's a huge augmentation potential there. So, you know, those are just a few examples, but I think almost every aspect of the insurance value chain, and I don't just mean between carrier and broker and reinsurer, but just within an insurer, I think there's room for greater efficiency for automation, for relieving some of the most taxed knowledge workers from doing the least, gratifying, least rewarding kind of work. There's enormous potential. There's this other universe as well, which is like policyholder service and AI contact centers and the like. I'm a little dubious about those, actually. That's just a personal opinion, because I don't think that those lead to good experience. I'm not yet persuaded that those lead to really high quality consumer experiences. And I think there's a lot of frustration just in the zeitgeist about those today. So if anything, I would be, if I were a McKinsey advisor again, which I'm not, but I would say, you know, just be careful how hard you lean into that, because there are just a certain category of problems that humans just need to talk to humans. And you can go too far and really estrange your customers. But that's just a personal opinion on that account. And I'm sure all of those solutions will continue to get better and more responsive, et cetera.
Juan de Castro: Marcus, this has been a phenomenal episode. I think we started with your experience at Guidewire, really interesting takeaways from that. And we moved on to your vision the areas where technology can drive value or insurance, and back to a lot of what you were discussing initially around why you started the company. So I think you're now basically investing in the areas where you think technology can solve a real problem with a clear ROI? And obviously we finished talking about LLMs. I think this is probably nowadays, and there's no better topic to wrap up the episode with. But yeah, all I have to say is thank you so much for joining me today. I thoroughly enjoyed the session.
Marcus Ryu: Yes. Thank you, Juan. I really appreciate it. And congratulations for all the progress your company is making.
Juan de Castro: Thank you. Making Risk Flow is brought to you by Cytora. If you enjoy this podcast, consider subscribing to Making Risk Flow in Apple Podcasts, Spotify, or wherever you get your podcasts, so you never miss an episode. To find out more about Cytora, visit cytora.com. Thanks for joining me. See you next time.
