Arik Yelovitch

Alright, so what do you do, sir?

I'm CTO and co-founder at Adaptive Insurance.

And what is Adaptive Insurance?

Adaptive Insurance is InsureTech, an MGA that is one leg in the insurance industry and one leg in the climate, so we are climate-oriented, InsureTech, I would call it. We develop insurance products that are helping to get the pain points of traditional insurance with parametric and non-parametric products related to weather. For instance, our first product that we went into market about half a year ago is what we call Grid Protect. Grid Protect is for short duration power outages for businesses. So any business that needs power, which is 99% of them, needs to operate,

and let's say you lose power because the grid went down and after eight hours you bought insurance for eight hours trigger and after eight hours we pay you the sum of the coverage that you bought and helps you to recover from wage loss, from food traffic that you didn't come into the door if you're a coffee shop or restaurant or any other thing. If you're a dentist, you lost all days molds and stuff like that, which you had to do in order to treat your customers or treat your patients. So we cover that gap, it doesn't count against you and it's a quick payout. It's not traditional insurance, it doesn't go for adjudication and stuff like that. So it's if else, if something happens, you get paid.

Interesting. So when you first started talking about it, I was like, "Oh, you can't get homeowners insurance in Florida anymore. Are you trying to solve that on the climate side?" But if you're talking about a power outage,

Austin storms a few years ago, weather goes out, we have no business, suddenly...

What happened now with the latest in January, that entire 90% of the state of Tennessee lost power. So that would be a huge get through us. We would think about it's a national disaster, yes, because it's a big event, but they would have received the first check after 24 to 72 hours from us and then they can deal with the regular insurance. It's not catastrophic, not the cat insurance, because in cat there are many more items.

But generally speaking, it covers the immediate loss that you have. If it's a big event that you lost everything, so it gives you some liquidity.

But it's more, as you said, for business interruption. For the grid goes down because of snow, because of ice, because of winds, the winds that we had this weekend in North Texas, the wind gas was 64 miles per hour in Amarillo and other places. So that's the first product. We have two more products which we went to market, which are called wind deductible, deductible buyback.

For wind and hail, we'll go through commercial and residential. So think about everybody now raises their deductible for wind. State fund does now raise from 1% to 2%, 2% for let's just say $500,000 house. That's a lot of money to pay out if you have a claim. It's not that everybody keeps there in a checking account, a few thousand dollars for claims. So we help to reduce that for a very attractive price. And if something happens, we cover whatever you bought down. So if you bought down from 2% to 1%, you bring in 1%, we're bringing the other 1% and then the insurance claim can start, because insurance will not pay until you pay the deductible. No matter if you'll say, I'll pay at the end or I'll pay in the middle, I'll definitely get the money

until you pay the deductibles, you won't get any claim process until that's met.

Was there anything like this in the market before?

For grid protect, no. Grid protect is the first one. There is in a BOP a 72 hours endorsement that if power goes more than 72 hours, but that's pure cat, catastrophic event, right? It doesn't go down more than 72 hours unless something really bad happened. But for short duration, no, there's nothing like that. We are backed up by Tokyo Marine, TMHC in Houston, Houston Casualty Corp, which is a very good partner for us. They have a lot of innovative products. I like to bring this kind of products to the market. So that's very unique and we see some traction growing with it. We're thinking also about kind of going to the homeowners. Now with homeowners, it's a bit different when you think about it, because homeowners, we can't really quantify the damage. If the power goes down in your home, fine. I'll leave the hour without internet. I might need to talk to my kids actually. The kids actually won't show up from the rooms, but there's no real substance damage. The damage starts when the power goes out for longer times. Now it's getting hot in Austin in August. I don't want to stay home. I want to go to a hotel or I have food spoilage, which is also something that a lot of insurance company removing from the policies. So it's kind of like, okay, low sums. Maybe I need $500, $1,000 coverage. Just we'll pay out more than 12 hours just to pay for a hotel room with my family or maybe restock the fridge because everything went bad.

So really it's a supplemental insurance to homeowners insurance.

It can be. It can be. It's not yet, but this is an angle we're working on and there's some traction there. For commercial, it's again, it's about business interruption. It's about how do we make you cope with business interruption better because businesses die by a thousand cuts. That's how it works. Now, most businesses don't just get wiped out of the map. It's one month after a month after a month of losses and eventually they close the doors. So this comes to help them mitigate something that they have to have in order to operate. You can operate without power. Even if you have a generator, which is amazing, you still have operational costs for the generator, right? Fuel costs money, diesel. And on top of it, if the entire Austin downtown is out of power and you're the only one with the lights on, nobody almost leaves the house when the power is out. Traffic lights don't work. You don't have people that are not out in the street when the power is off. They stay usually in their houses or in a safe place relatively, and watching for their own stuff. So regardless if you are well suited for it or not, you will have business interruption.

So it's really climate resilience.

It's exactly climate resilience. And that's one aspect. And that's why I said we are one leg in the insurance industry, which is the insurance products. But the other leg is exactly what you said, the climate resiliency. We developed in-house a weather platform,

which calculates tons of data. We actually quadrupled our storage for data in the last two months. So we collect weather information. We collect all the power outages in the U.S. We came up with a score for power and a score for weather. So we're able to assess for every given address in the U.S. We're able to assess the risk of power loss and weather phenomenons. And we're able to predict some of the weather with some of our partners, which we use. So we kind of try to use a lot of analytics, a lot of AI to analyze. There's a lot of data out there. It's a matter of analyzing it. And AI allows us to analyze it. So that's the climate aspect of us. We're able to talk to our customers and say, "Hey, we see by this pattern, ABC will happen. You're on a flat plane. It didn't happen for 50 years." But predictions say it might. So you should probably move your business. You don't build here, build here. It's much safer for you, et cetera, et cetera. So we're able to monitor life, a lot of stuff, and analyze it, and for better pricing, for better underwriting on the insurance arm, but also to help customers to kind of prepare and understand what they're going to face. Because weather, regardless of, you know, climate change, no climate change, why it happens, that's a political argument. Something is happening, right? The weather is more, less predictable and more extreme.

We see that, like, between Sunday and Monday, we have a drop of 60 degrees Fahrenheit. It's just, I don't think I ever experienced that. Unless you jump on a plane, fly somewhere, and land in the next morning. Same place, 60 degrees, that's extreme.

It was hotter in San Francisco when I was there than it was in Austin.

Yeah.

It was very strange.

It was colder here. We were in London at the end of January. We just missed because we had an important meeting, so we kind of felt like, don't quote me on that, but someone told us, "Do you pull a Ted Cruz?" I said, "No," because actually my family was still in Austin, so I didn't. But it was hotter in London by, I think, about 40 degrees Fahrenheit, than actually in Texas at that week, at the freeze weekend, at the end of January. So it's crazy.

That is crazy. All right. So the other topic that we're talking to a lot of people about is, you know, you mentioned that you're using AI for pretty good models. Yeah. And obviously, how long have you guys been around?

I've been in a company for almost two years. In a month, it will be two years. All right. And the company is, my co-founder started it in February, in January and February, and so two years and change.

Great. I mean, the models keep getting better.

Oh, it's crazy. The controls keep getting better. It's crazy.

So, you know, you're teaching at home. You've been playing around with software for a while. What are you seeing out there right now that's got you excited?

So the two aspects for AI, one is the analytical aspect, the data aspect, and the other one is what I call the development aspect. Right. Currently, AI changes the game from development perspective. I don't like it. Some companies take it further than others, and we see what happens due to that. And I don't think it's mature enough to let it go and ride by itself. So AI, I think for now in development, is a huge, huge, huge accelerator. Things that would take me just from pure, you know, time consuming. Writing code takes time, regardless if you think about it or not. You type X letters in a second, right? It is what it is. It's physics. It allows you to generate a lot of code in a very short time span and smart code. Not necessarily good code, but in a like human readable. But what we, I heard the term and I really liked it. A lot of people now code side by side with AI, which is amazing experience. I do it a lot. It helps. It's a feature that would take me three, four days. It takes me a day because pure, it writes everything. And then I need to go in,

analyze it a bit, review it, correct it because we are in the insurance industry. So we're highly compliant and AI doesn't understand that niche yet. Up to, we are now developing, we are developing our own LLM that just put a ticket, simple ticket, will just run by itself, create a PR, run test.

And in the future, probably won't also, you know, push the production. I'm not there yet. We saw what happened with AWS three months ago, four months ago, GitHub. They all went too far in my opinion.

But it's a great accelerator for developers. The only caveat is, and I told it to my friend, it's like Coble developers in, when I started my career in the 2000s, everybody says there are no more Coble developers. There's no, nobody studies Coble anymore. Coble is a dead language. You know, they say Coble developers only die because mainframes is the past. Web is the future.

Five years later, you see ads in the newspapers. Coble course, please join paying $5,000 for you to learn Coble.

We don't shift that fast. So with AI, everybody tried to say, we won't hire anybody. We won't have any junior engineers anymore. It doesn't work like that. But even if it does, my concern about that is that in order to be a senior engineer, you need to be a junior engineer. And we create it like in the medicine, in anesthesia and other domains in medicine, you don't have the next generation. Now maybe it will be kind of redundant. We'll be AI developers or AI helpers. But for now, I think that's a kind of double edged sword. So you have to balance it out. Yes, it will save money on developers. I don't need as many juniors as I should have in my scale. So we are a very small team, but you need to understand that it's not replacing. It's an accelerator and you still need a human touch there.

How big is your team?

We are one FT engineer and four contractors. So we are about five, six people.

Are you contracting on Tel Aviv?

No, we have contractors from Brazil and we brought someone with H1B right before the price went to 100,000 K. So he actually dodged that bullet. He was like, we said, let's submit you before the president actually raised the price. And he got selected. So he said, you're lucky because next year you we would not submit you 100,000 dollars for a startup is a bit expensive. So we're a small team, not many too high up. I was planning to hire and then we went deeper into AI. So yes, that's I'm following suit and saving money on developers. But again, it's a balance that I have to keep regarding data. There was a lot of data. But problem with a lot of data is that there's not enough bodies you can throw on the problem to solve it in a normal in something that you can actually build upon. Right. It can take years. But if you analyze data from now and it takes you two years to analyze the data, then this is a two years old data. It doesn't help because the world moves so fast. AI enables us to basically take this terabytes and I don't know what the next unit everybody in that Google bites or whatever the peta and whatever names will invent very, very soon and analyze it and come with conclusions. And it's getting better and better at it. It's less hallucinating because we introduce also tools to help it. I read and I was in a lecture that says AI starts to hallucinate when it doesn't have tools, analytical tools, and it just starts to make assumption.

And we all know what assumption is. And in AI, they call it hallucination. So introducing also tools to the AI to be able to come with an empiric decisions and not just, hey, I think it's work like that. And let's make it so. So it allows us to move forward really fast with analytical and the analytical platform we have, the weather analytics, and it's just exciting. Again, if I look at the data team that to develop, we have our head of data, which is an amazing, amazing woman. She's actually moving to Austin now and her engineer, we don't need many more. We need one more climatologist or something because it's still knowledge.

But we don't need worker bees on that because AI does so much.

So on the other side, what are you seeing right now that's keeping you up at night?

What am I seeing now that keeping up at night?

Not a lot of things. And mainly the success of the company, right? You can fit to market. I think fit to market makes me parametric. Parametric insurance was is here for a long time. Nobody was able to solve how to sell it in the mainstream. It's a luxury product. How do we make it so businesses and homeowners will see the integral part of their insurance portfolio, of their safety, of the safety net? How do we make it part of the safety net for them? That gives me up on night because at the end of the day, we have great ideas. If you don't sell,

that's what 80% of the start of the fail didn't do. They didn't sell.

Getting there and finding the right ideas and pivot fast. And we pivoted so many times. Pivot fast if we don't feel it's a fit. Be able to admit that it's not the right fit, although it hurts your pride because we always have the most amazing ideas, of course. But sometimes they just don't fit reality.

That's what gives me a night. Otherwise,

just me pushing forward.

What tools are you using right now? We

use cloud, we use Gemini, mainly cloud and Gemini. We don't use much at GPT.

We started developing some bots on top of cloud. It's a great platform.

I think they're a bit overwhelmed, so the customer service is a bit lacking. But that's okay, it's a good problem to have.

But yeah, we're not going to reinvent the wheel. Cloud and Gemini are amazing platforms.

Are you playing with OpenClaw or the new Nvidia version of it?

And not yet. We're still in the development side. We still see it and respect it, but still trust, but verify the phase.

But yeah, once we get it mainstream, we'll start moving forward. It's also pricing for now, right? We're a startup. I don't want to put pricing as a barrier or a limitation on AI, but I'm not going to go to the latest and greatest. I don't necessarily need it. Also, in development, you don't always jump to the latest version. You let it sit for a while for others to get all the bugs out of it.

We're a staffing company, like we hire resilience for other people.

One question I'm constantly asking people with the AI side of things is, let's say you find your market fit, you get your traction, it's time to scale. Do you even see the need?

You see the need for hiring, but what kind of roles are you hiring for in the future? What does that look like?

In the engineering side?

Yeah, on the engineering side, with AI.

So QA, I still think it's important. Human Touch, we hired a part-time, like for our contractors, they gave us a part-time QA, but I want to have someone in the US sitting shoulder to shoulder, because I think QA is still more than just an engineering position. I'll give you an example at Hippo, which basically the company that brought me here to the US, we had an amazing, amazing QA. Her name was Xenia, still Xenia, but when she worked with us.

She did much more than just QA. QA have to learn the business, and if there is, again, specifically and mostly important in insurance, because it's much more than just a SaaS product,

and she learned the business and found bugs that developers wouldn't be able to find, because we don't necessarily understand the business logic. We don't have time. A developer, like a line developer, whatever we'll call, doesn't have time to go into the needs and greets of the insurance. It gets a spec from PM, which hopefully knows insurance, and implement it, and you have to have someone who's knowledgeable. I think QA is something that I will definitely hire as an FTP.

Engineer wise, yes, at some point I will have nest need for junior engineers, and I will hire probably another senior engineer.

That's kind of, I'm playing the game, unfortunately, because getting now a junior engineer in AI,

unfortunately, and I'm saying it with great salt, because my kid wants to go learn computer science, and I'm like, I'm not sure you need to. I'm not sure that's the right path right now. Luckily, he's a junior, so he still has time.

That's probably the hiring. More senior people rather than juniors, but on the QA and engineer side, and of course, a technical product manager. That's something that is going to come very soon.

Six months from now, who knows what anything looks like? These are really so quickly.

Yeah, they're moving so quickly. They're moving too fast. Again, I'm on the cautious side with AI. I know companies, as I said, are very declared. We're firing half our department because AI replaces them, and then you have AI push a score, and you're asking, "Why did you do it? Because I thought so. Will you do it again?" Yes, you would fire any employee that would say so. Again, we're just in the start of this journey, the beginning of this journey, but I think people are jumping the gun, and the engineering industry and tech industry is known to jump the gun. We did it a lot. On a lot of occasions, the new technology came in and we're like, "Yeah, this is the best thing since the wheel, since they invented the wheel." Then we're like, "Oh, maybe it's not. Maybe we still need to wait and see, and tweak it a bit before it will become perfect as we think it is."

Well, hopefully it won't crash the entire world economy.

Listen, if AWS crashes again for three days, we actually might. Who knows?

But that's not keeping you up at night.

No, that's not. Let's say if AWS crashes have bigger problems than adaptive.

When we started at HPO in 2017, AWS crashed for the first time, the entire eastern region for, I think it was half a day, and the CEO came in, "Why are appies not up?" Like we were on the cloud, of course. He said, "I suck, really?

You think this is the biggest problem right now in the world?" We just started with HPO. We sold maybe two, three pauses a day. You think this is the biggest problem now that you're down?

A lot of credit card companies are down. A lot of SaaS products are down. People are not worried about the insurance right now. They're worrying about access.

So we're the least of the problems for now.

Well, it's interesting watching servers become true utilities.

It's essential infrastructure. It breaks the planet right now, and it's not working.

Internet and power became essential. You can't live without it anymore. Try living without internet for a day, unless you're camping and you do it on purpose. Even on camping, I see people taking the phones, trying to find some reception and see what's going on in the world. We became addicts. It used to be news flashes every hour to hear the news. Now we're addicts, and I'm one of them. I'm no different than anybody else.