Building with Corgi - Kanyi Maqubela in conversation with Nico Laqua & Emily Yuan

I'm Kanyi Maqubela from Kindred Ventures, and I'm joined here with Emilie Yuan and Nico Lacqua, the founders of Corgi Insurance Services. Thank you guys for being here with us.

Thanks so much for having us.

Yeah, excited to be here.

Okay, let's just dive right into this. So you market yourselves as AI insurance. What does that mean? What is Corgi?

The way that I describe Corgi is that we're really interested in the most regulated areas of the economy where a lot of money flows kind of in and around. And what we found is that these super regulated entities, something that's unique about them is that the data primitive that they use is language and tends to be language. And something really exciting about computing is that computers have very recently gotten good at using language. Insurance is the largest single part or area of financial services. It's about as big as payment processing, credit cards, and banking combined. What we try to do is run kind of these highly regulated financial entities, particularly insurance companies, particularly insurance infrastructure without people by having large language models communicate with other large language models where possible.

You said that insurance is bigger than payment processing, banking, and credit cards combined. Is that because of all the reserves that you have to keep on insurance, or is that on as a function of revenue or as a function of market cap? Because it actually almost doesn't even sound real.

Yeah, I mean, insurance, it definitely isn't an area where a lot of startups have tried to do things. And there's a couple of reasons why. The main one is technological. You know, it's a bit different than like if we talk about payment processing, banking, and credit cards, the way that they all kind of work is that there's a general ledger, right? Like a database or a spreadsheet that processes a lot of numbers, a lot of payments, normally very quickly, ideally an increasing amount as time goes on. And normally a cut of all those type of transactions is taken by the infrastructure provider. Insurance isn't quite like that. So in insurance, what we sell is called a policy. It's a contract, it's 100 pages long, and turning that contract into math is the work that's done. So there's a lot of very large insurance companies. I can name off the top of my head more than 20 of them on public markets around the world that are worth more than $100 billion in terms of market capitalization. But something that all of these companies have in common is that they tend to be pretty old school. Think like '80s back office, fax machine, phone calls, all the bad things. They tend to run without a lot of technology. And most importantly, there's a huge regulatory and regulatory capital barrier to entry. So insurance is a really important part of the economy. It's about 12% of the GDP, about twice as big as software, but it's just kind of an untapped area of financial services in terms of startups because it's particularly language-heavy instead of numbers-heavy.

So insurtech had its moment in the sun in the tech industry over the last 15 years. About 10 years ago, it really, really became hot. Are you an insurtech? And if not, what's the difference between an insurtech and an insurance company?

Yeah, I don't, I don't like using that term. And the reason why is because kind of the last generation of insurtech, there's some good companies that got liquidity and ended up doing well. But by and large, most of the focus was on kind of the distribution layer, being a broker or an agent or an MGA. Which are all pretty similar things, but basically those are companies that resell someone else's products. And the type of entities who are making the products that they're reselling are insurance carriers and reinsurance companies ultimately hold the balance sheet risk. And I refer to those as insurance infrastructure or financial infrastructure companies. We're kind of in that second category, which is somewhere where we haven't really seen startups in the past. And we're not a consumer-facing company at all. We have no consumer-facing products. What we do is we price policies, we hold balance sheet risk, we communicate directly with regulators as one of our key stakeholders. We try to go down to the regulatory bare metal where we can, which is super different than— I don't know if I should name them, but then being a company that's taking 5 or 10 or 15% off the top or jacking up the prices of someone else's policy and selling it to someone else.

When you guys pitched Kindred Ventures for your seed round, you said that you wanted to buy an insurance carrier. I didn't fully grok what that meant because you were raising a pretty small amount of money. And my understanding at the time was that insurance carriers, which is, you know, these underlying regulatory attached building blocks that power this industry, really expensive and it was extraordinarily capital intensive and capital consumptive because you've got to hold the reserves and there's all these regulatory requirements around cash and capital on your balance sheet that you need to keep up with. How did you raise a seed round to buy a carrier? Can you walk through the TikTok of that?

It definitely wasn't an easy thing to do or that pleasant of a process. It ended up taking us about a year and a half. To get all the regulatory approvals. And the seed round that we raised didn't actually buy us the carrier. What it got us was a foot in the door and a call option with no ownership on an insurance carrier.

So thank you, Kindred, for taking the bet on us.

Yeah, so we raised $6 million on very short notice. And I think our plan was— I think I can probably say now our plan was a little more half-baked than I would have liked it to, although it came together over time. Our go-to-market, because you can't really make an insurance carrier or kind of insurance infrastructure company, necessitated us actually acquiring underlying regulatory licenses in order to make that kind of company possible. The amount of time that it takes to make one of these highly regulated financial entities is pretty extreme, and the amount of reserves and experience in insurance you need is also pretty extreme. So much like Column and other kind of financial infrastructure companies have done in the past, our hack was actually acquiring that insurance company. The end price ended up being about $35 million. And it took us quite a long time to get all of the regulatory approvals and prove to our regulatory stakeholders that we're credible, we're liquid, we know how to actually run an insurance company, all of those sort of things. In order to get it done, we had to, you know, kind of earn that trust over a a long period of time working hand in hand with the original management of the carrier. Yeah, it's definitely not a very easy process or one that looking back I would recommend.

Is $35 million a good price for an insurance carrier? Like there's not like a rich M&A market for it or anything like that.

It's a good price for us for sure. Because otherwise I think it would've been, impossible to like even play ball or enter the area. It's not— I don't think it's the kind of business where like a private equity firm would have been very interested in going into the market, especially because of the kind of long time horizon in order to become an insurance infrastructure provider. There's, for reference, within the type of insurance infrastructure company that we ended up acquiring. There's about 200 that exist in the United States and we own 2 of them. So we own about 1% of the total insurance carriers in our category. And I think that is really invaluable to actually, because core to our thesis is that if we see a super highly regulated entity where a lot of money flows in or around, or it's a really important part of the economy, we want to actually be that entity. We don't want to sell to that entity. We don't want to partner with that entity. We don't want to be a wrapper on that entity. We don't want to actually be the highly regulated financial entity. And in order to do that, you know, kind of the, the list, if you might imagine, of potential insurance carriers or reinsurance companies is, is, is pretty short. So for us, it was, it was a phenomenal price, although I think raising at a, at a $28 million valuation to buy something worth $35 million, you know, the math is a little tough. All right.

Let's talk about technology. Well, you studied computer science, uh, your time at Stanford. And, uh, obviously it being an AI insurance company in this era means that you're working with large language models. You're dealing in one of the most valuable language-based, uh, industries in the world. Can you talk a little bit about the stack and what makes your stack and maybe even the steps, sort of speaking in insurance terms? So I guess claims handling and and third-party administration and underwriting, how you're able to streamline those things with, with, with AI.

Yeah, so insurance is made out of a lot of kind of these departments that each do some amount of document processing, whether it's reviewing a claim that comes in and reviewing receipts to determine the amount that we should pay out for the claim or administering the policy, getting into the policy administration system. To be able to read and process a lot of documents. I think it's just very important for us to do the automation that we're doing right now. And previously, like, that would have been completely impossible.

Do you have to train a model or can you just plug these documents into an open source instance of an LLM and it just runs, or?

We don't have our own trained model. We just plug it into the LLM. But there is a lot of, it's less about the model in our case and more about the specific workflows. The workflows of like exactly how you need to process the document, how you need to like tie the different inputs back to the insurance policy, it's like very, very nuanced. And being able to like map that workflow precisely, that's where there's a lot of work that needs to be done because it's very complicated, very regulated, and you can't mess up.

Now, I would imagine that what you're describing there with these workflow automations and presumably some supervisory models that are building intelligence on how to optimize for different regulatory paper could be sold to other insurance companies. And so as you think about where you're generating the most value and maybe even where you're going to market with the most velocity, uh, what is the argument for verticalizing and doing it on all yourself rather than just becoming a SaaS and selling this AI workflow automation capability across the industry?

It's hard to just have one piece very like, oh, just like plug this one automation into the otherwise like very archaic infrastructure. So what we had to do is kind of build all the workflows from like an AI-first, like automation-first point of view. Otherwise you're trying to like going backwards, like plug this technology into like these very like human-centric systems. And I think that makes it make less sense.

The, the, the other thing is that insurance is a fundamentally good business as long as you don't get greedy and you don't get stupid. You know, many of the biggest companies in the world are insurance companies. There's a lot more super big insurance companies than super big SaaS companies. And just fundamentally a bigger portion of the GDP is insurance. More profit is made, more revenue is made than software. So I think we're kind of a bet in many ways on models getting a lot smarter and the models have gotten a lot smarter and a lot better, which is why a big part of our philosophy around building versus using existing, you know, we used to do all sorts of RAG and pre-training stuff, but now, you know, context windows are huge. The models are a lot smarter. So I think what our company is a bet on is that ultimately these kind of highly repetitive, highly regulated tasks that normally require tens of thousands of people at scale can instead be done with practically no people, maybe just a couple of really smart engineers. And that will lead to better outcomes for end customers. That will lead to better outcomes for our regulatory partners. And that will lead to faster, better, and ultimately cheaper insurance. And that's that, you know, it's not like the most exciting thing that everyone's waking up in the morning raving about. But when you're talking about 12% of the US GDP, it becomes very meaningful. It's like having gasoline that costs $0.20 a gallon instead of the very expensive price that you might see at the pump right now. Immediately everyone would notice, I guess, that something like that would happen. But there'd be enormous downstream consequences for the whole economy. Basically, this industry that in many ways is, is a necessary part of doing business— without it, every single business would collapse— becomes less of a tax on the economy and becomes much more efficient. So I think that like while maybe a great company could be made selling to these very large enterprises, actually becoming the large enterprise is something that's more exciting, more interesting, and does a lot more for the world, I guess, in a similar way that maybe Uber could have been a phenomenal business selling all sorts of vertical SaaS software to taxis. But I think that making a new type of transportation system made it fundamentally a more exciting and interesting type of business.

It's funny when, when, uh, on the original slide deck for Uber, there was a TAM slide and the TAM slide was orders of magnitude too small because by creating a new transportation category, they weren't just able to capture the taxi market, but they were able to take a piece of transportation itself. So it sounds like that's what you're getting at with this argument. All right, let's talk a little bit about your product lines. So what are you actually insuring or what are your customers insuring and why are those the places that you've started?

So we started working mostly with property managers in our vertical real estate market segment. Our big initial kind of suite of products focus on on resident and tenant compliance, ensuring that people have coverage for our alternative renters product, our security deposit product, and our pet liability product. The reason we chose to start there is because there's no real potential for catastrophic losses. The policy limits are relatively low, and the margins end up looking a lot like software margins that people in our industry are much more familiar with. Our next big expansion that we're really excited about is our biggest expansion is providing insurance for technology companies. We're both directly insuring these technology companies, providing them D&O in case their investors or someone else sues them, providing E&O to similarly cover them if something happens. Providing CGL so they can get a lease and get some protection there, cyber, commercial auto, all kind of these categories that technology companies need to function, as well as providing insurance for these technology companies to operate kind of as an insurance infrastructure as a service provider, wherein these companies can sell insurance to their end customers. And for most vertical SaaS and vertical agent companies, that's particularly exciting because in general, insurance is just a much bigger kind of area of spend than software. It's a much bigger part of the economy. Most businesses spend a whole lot more on insurance than they do on software, normally about twice as much, sometimes more than twice as much.

Is that true of startups though? Because I would imagine that a startup is spending most of their P&L on software and headcount, right? So Is that true of startups too?

Startups are kind of a category of business that I think spends relatively little on insurance, but it's still pretty meaningful. I think most startups of our size are spending at least $50,000 a year on insurance, something in that ballpark, probably roughly on par or maybe a little bit less than our software spend. Most businesses in general are spending a lot more on insurance than they are on technologies.

Follow-up question to that. So for the average startup, maybe we'll take Corgi as an example. You know, you've got all these different insurances. You have commercial liability, you've got insurance on your office space, you've got insurance for your executives, your board of directors. Is that across, like with each one of those lines, is it 5 different insurance carriers that is delivering you each of those 5 products? Or is it typically one carrier? And in your vision, are you suggesting that Corgi is going to be one carrier that's going to consolidate that, or are you going to be one among the five?

Yeah. So right now there's two companies mostly that service a good portion of, of technology companies. And what these companies are, are brokers. And what brokers do is they take policies normally from a carrier called State National, which is this multibillion-dollar insurance carrier. I promise you they don't care very much about startups, so I don't mind naming them. And, and basically they'll take these state national policies and they'll jack the rate up a little bit and then take a cut. Normally it's between 5% and 15%, but it can be anything. And they'll resell these state national policies. So ultimately, most startups right now are buying policies from a broker that has no control over the product quality, over the pricing, over whether things are covered or not. Most Most people that run startups right now probably are reluctant to file claims because they assume that they won't be taken care of. And I think unfortunately they're probably pretty correct to be worried about that. We're a bit different. We don't have any middlemen at all. There's no brokers selling our stuff to technology companies. We are the carrier. If something goes wrong, people can text me and we provide coverage ourselves. We sell products that sit on our own balance sheet where we deeply care about the quality of the products that we're selling. We have full control all the way down to the regulatory bare metal, including the reinsurance.

And I think another note on the product side is we are able to customize these policies a lot more for what startups actually need versus just having a super generic general liability policy where, oh, you're like a, You're a startup in an office building, you don't have a roof, but it still covers your roof, right? You're still paying for that roof coverage. We can add in things like covering for like AI malpractice, which is pretty rare, and having it be like very tailored towards companies that sell these AI services. And yeah, just being able to design the products around tech companies and not just taking some generic template and trying to retrofit it into like, oh, let's take these 5 things out and then we'll add these other random 5 things. And okay, now it's a customized package for tech companies, which is what happens when you use one of these. Yeah, a lot of these other—

if you're using a broker, normally there's quite a few middlemen before people are actually getting their product. So you have the regulators and they're a very important bunch. But on top of the regulators, you normally have reinsurance and then you have an insurance carrier, then often MGA. Then on top of the MGA, you have a wholesale broker and then a retail broker. And that's the person you're interfacing with. When you're working with Corgi and you're getting your technology company covered, you have Corgi and then I guess the regulators and that's it. So if something is going wrong, you can reach out to us., and we'll make sure that people get taken care of. And most importantly, if you're interested in selling insurance to your customers or providing insurance as an embedded service that your customers might be interested in, then we can make sure to enter conversations to facilitate that as well.

So in terms of this customization that you're talking about, I want to drill a little bit more deeply into what that means because some companies have to get general liability for their roof, whether or not they have a roof. Does it even go down to the granular level of what kind of products they're specifically selling? And that sounds like it's pretty time-intensive. Are you able to automate that as well?

Right now it's customized for tech companies in particular, but going forward, like we are able to, because we are the carrier, we're able to add in a lot of granularity whether they are a SaaS company, whether they are like hard tech, whether they're like a robotics company, like all these different types of tech companies, we get very granular in like the actual coverage that they need and just provide them the coverage they need and leave out everything that they don't. So they don't have to pay for everything.

You mentioned AI malpractice. Yeah. First of all, it's interesting that you use the phrase malpractice because I typically think of doctors. And so I want to hear why you frame it that way. And then second, is that a fast-growing area? Can you articulate maybe where that's showing up thus far? And I can imagine if eventually all these technology companies just become one-employee or zero-employee companies, that AI malpractice might be the whole thing that they deliver at some point. So what's going on there? Can you say more about that?

It's very much like a recently emerging risk category where there's a couple industries where you have very common malpractice professional liability type policies, and those are— you have like doctors, doctor malpractice, medical malpractice insurance. You have lawyer like malpractice and like liability insurance. And then you have tech E&O insurance, which is kind of similar to those where like, hey, like if you implement your tech and it goes and it messes up and then your customer loses like a million dollars in revenue, and that's what it will cover. But sometimes those carve-out cases like sometimes it just carves out like AI use cases. Our product allows these, hey, like if you're in, I don't know, like AI, like phone answering company and you work with small businesses to help them book meetings and book appointments, and if it goes down and then they like lose a bunch of revenue because your system glitches, like that's something that we could cover.

I suppose what's interesting about that is system glitches have been a problem with technology for the history of technology. But malpractice, especially when you compare it against medical malpractice and legal malpractice, sounds like human error and it's sort of like covering human error. And so it's interesting that you call it AI malpractice because it's artificial human error, sort of like what you're getting at. So yeah, I'm intrigued by the use of that phrase.

Yeah, the things that could go wrong are a lot more human-like.

Right.

Like when a business uses AI technology, they're using it to replace some, like, yeah, some previously more complicated, like human-involved workflow. Whereas usually, like beforehand with technology, it's more like, hey, there's like this set one, like this one small set process and it just keeps automating that. Or it's like a CRM or something.

So it either worked or didn't work.

Okay, so either like it worked or didn't work. In this case, like, There are a lot of ways where AI could mess up.

So something really interesting about insurance in general is if you look at the history of insurance, you know, it's older than banking or money itself. And it's one of the, and when you have any financial economy, insurance is one of the first things that, that like must arise. So if you go all the way back to the Bronze Age when they were shipping things in, in North Africa and, and the Eastern Mediterranean, if I have a boat and I'm shipping things and pirates take all my stuff or a storm sinks my boat, all go out of business without insurance. So spreading that risk is something that's super necessary. Otherwise you can't have a functional economy. And that's kind of a core service that's necessary in order to have any financial economy. Otherwise, every single business has to overcapitalize. They have to have enough reserves on hand to support any sort of, you know, disaster that might happen. And that's just not economically efficient for that, that for for every business to have to have a ton of cash sitting in the bank that they can't spend on anything.

It's better for a system to socialize the just-in-case risk.

Yes. And we're particularly interested in a lot of ways in kind of frontier industries. These frontier industries, in order for them to transition from these venture-backed models where companies need to have a lot of money sitting in the bank to cover the just-in-cases, like in order for them to become sustainable kind of areas of the economy, insurance is a fundamental necessity. So there's a lot of these frontier industries that are really interesting to us in the future. Like space is another example.

All right, let's talk about culture. Corgi is famous. I think the Wall Street Journal or another national news organization covered how you can earn a mattress after your first month. At the company, that you're not a 996, you're a 997, that you've got this super hardcore culture. On the one hand, it feels like a sort of modern trend that's sort of important to the AI era and important to this opportunity at hand, which is I think what that article was talking about. But on the other hand, that's how high-growth startups I think have always been. Do you feel like you're breaking new ground or do you feel like this is a return to how things used to be? How do you guys, if you took yourself out of your fishbowl, look at your own culture?

I think good startups have always worked really hard. There's like a lot of famous stories with like Bill Gates, like having— what was it like the 12-hour turnaround or something?

I thought it was like 7 hours.

7-hour turnaround. Okay. Yeah. And that's pretty cool. I think symbols really matter and our company tends to consist of immensely entrepreneurial people. We have an enormous number of of ex-founders on our team, used to be the majority, and now it's still pretty close to that. If you're not working 7 days a week, you're kind of quiet quitting a little bit. Like you're, you're not really giving it your all just because when you're writing code, when you're working out deals, when you're, you know, coming up with new ideas, I think that, you know, certainly putting in that incremental extra effort is, is something that's kind of required in order to win, in order to be high crow. In order to create something that's actually important during this technological shift.

Do you worry about burning people out?

Most of the people on our team, I'll say this, have been working 7 days a week for years before we came in the picture, and that's true of Emily and myself as well. So it's not for everyone, and if you value work-life balance a lot and you don't want to work very hard, I can recommend a lot of companies to you. Corgi is not one of them. And I think that because we have a really strong sense of community, we spend a lot of time together. Many of us go to the gym together, we eat meals together. We had a lovely Thanksgiving dinner for those that were in the office with me. And but this year and last year, for all the major holidays, I'm there. Corgi is a really unique environment because there's a super strong sense of community amongst everyone there. So I don't think a lot of members of the team have burned out or will burn out, but it's certainly not for everyone. And a lot of our team members have families and, and obviously there's, there's sacrifices one way or another that people need to make. But if you want to win and you want to make a really important company that people are talking about for the next 100 years, then you're going to need to put in some work.

This is the AI era, and one of the defining characteristics of it is these super steep high-growth cycles of adoption. And some of them, they're empty calories, and so it's just as steep of an adoption as it is as steep of a churn. Can we talk a little bit about both, I guess, adoption steepness and churn in a business that's older than banking, right? Like how has that been? Taking something as old as insurance and as new as the sort of AI culture that's super intense and super high velocity and smashing them together, which one wins maybe is one way to ask the question.

I think insurance is fundamentally, it's a very sticky business because even if your company's going out of business, like the last thing you cancel is your insurance. Like everyone needs insurance. Right now it's just a huge pain to like switch insurance companies, to find insurance. It's a long process and it's just something that you need to like check a box, get done and kind of just, have it be done. You don't want to worry about it. So because of that, like, churn is really low. It's not like SaaS companies where you have people like try something for a month or two to see like, oh, like, let me play with this new AI tool and see like if it can make some stuff. And then they like, they play around with it for a little bit and then they decide like if or if not they want to use it. Like for insurance, like you need to have it.

Really low, like 10%? Really low, like 5%? Really low, What order of magnitude, let's just say?

Really like sub-1%. Wow. Yeah.

You know, we've been growing pretty quick. It turns out when you don't waste people's time and provide them the financial services they need, they want to sign up. And for our embedded partners, we make them a lot of money too, which helps. We're definitely one of the faster growing, I don't know about the fastest, but definitely one of the faster growing B2B companies of all time. By revenue. We hope that that continues.

Thank you guys so much. It's really, really great to get to do a little reflection here. And I know we're at the beginning of this journey, but you've already accomplished a lot. So congratulations and thank you for spending time with us.