Attention all homeowners! Answer the following questions: What is your roof made out of? How steep is it? What is your home’s siding made out of? How close is the nearest fire station?
These are common questions that homeowners must answer in order to get a proper quote from insurance companies. But what if they don’t know the answers?
Well, they wouldn’t be alone.
“I’ve made up answers to those questions several times over the course of my lifetime, because who the heck knows,” Ben Tuttle, CTO of Arturo, said. “And the person on the phone with you doesn’t know either.”
If the homeowner doesn’t know the answer, the insurance company might send a person to your home to find out — but that’s costly, time-consuming and also includes the potential for human error. Or maybe an insurance company will rely on public records, but those records might not be accessible or up to date.
That’s where Arturo comes in. The AI property analytics company helps insurance carriers improve the accuracy and speed of decision-making across claims, underwriting, pricing and renewals by delivering on-demand property data. And that’s no exaggeration.
“It takes us about five seconds to generate that data,” John-Isaac Clark (jC), Arturo’s CEO, said.
That’s because the company uses a configurable API and utilizes multi-source imagery, which provides detailed insights into the physical world at high accuracy and speed.
“And thus, the insurance agent on the phone now knows the answers that you don’t,” Tuttle added.
By leaning heavily on AI and machine learning, Arturo has entered a new frontier in geospatial imaging as it sets out to create the greatest understanding of places and spaces. Earlier this year, the startup secured $25 million in Series B funding, and since March 2020, its headcount has more than tripled.
Built In Colorado sat down with the CEO, CTO and lead geospatial data engineer to learn more about what makes this corner of the market a unique business opportunity, and how Arturo has been able to disrupt insurtech so quickly.
How has geospatial imaging evolved over time?
Ben Tuttle, CTO: People have been extracting information from geographic data for hundreds of years. The context for how people do it has changed. Back in the ‘70s and early ‘80s, a satellite would drop a film canister that got caught in a net by a plane. And then people would put that piece of film on a light table and stare at it. In the late ‘80s and early ‘90s, people would stand around a table with an image printed on a piece of paper, then draw on it with Sharpies.
John-Isaac Clark (jC), CEO: When Ben and I started our careers in geospatial imaging, an airplane or a satellite would go over a place and it would create this massive image. We’re talking gigabytes. A single image was on average eight gigabytes for just the visible spectrum — the pixels you can see with your human eye. Then you’d have to download it to a desktop, and with bandwidth speed, that would take about four hours. Once it downloaded, you’d need a highly specialized workstation that used advanced tools to open a file that big. Then you would have to zoom in on the area, and try to decipher if something was a roof or not. Just five years ago, to do what we do at Arturo in five seconds, it would take an analyst probably the better part of a day or two, maybe even longer, to get the data.
Jeff Albrecht, lead geospatial data engineer: jC just described what we call the desktop GIS era. That’s when you had these massive payloads coming in, then would use proprietary softwares and pay 200 people who have master's degrees in remote sensing to work on a 24-hour, three-day shift rotation to analyze imagery all day. None of that is scalable, because it just takes so long to get insights to the customer. That’s the end goal. Imagery data is cool, but it’s relatively useless to insurance companies in bulk form. They want insights that we pull out of the data.
Tuttle: The world has changed very drastically in the last 40 years as far as this stuff goes. AI is an interesting use for this application because, while there are a lot of analysts on the ground and the human brain is a magical thing that can do a lot of stuff, the brain is not super efficient at identifying objects in images. Computers are really good at that. People are good at telling you why that matters.
In a lot of ways, what we are doing is elevating the ability for those people, whether they’re an insurance analyst, an intel analyst or a financial analyst, to look at the data that comes out of our technology and ask themselves, “Do I care about this change that happened?” And, “Do I need to make a decision because of it?”
How does Arturo’s tech change the game when it comes to the insurance provider industry?
Albrecht: Normally, an insurance company will hire someone and pay them to drive their truck out to a property and get up on a roof. Or they ask questions of the consumer, or they look at public records, which becomes really difficult because every county in every state stores that data differently. Sometimes the data isn’t available, so they would have to page through tons of data. Neither approach is sustainable, and neither is particularly cost-efficient either.
Imagery data is cool, but it’s relatively useless to insurance companies in bulk form. They want insights that we pull out of the data.’’
Clark: Everything Jeff said is correct — and it’s no wonder that some of that data collected is wrong or inaccurate over time. Because change happens and humans make mistakes. These methods are costly, time-consuming and expensive. Arturo changes the game in three ways.
One is accuracy. Any time a new image is available for a property, we can analyze it, and we have fresh, accurate information from that image. Two is time. We can do that in about five seconds — faster when we're doing many properties at once. It doesn’t take a week to send someone hired by the insurance company to inspect and write their report that answers the questions insurance companies want, wait for them to send it back, and pay them. And three is cost. Information economics changes drastically when we can analyze with machine learning and AI instead of humans.
Tuttle: The parcel data that Jeff is talking about that the counties aggregate — typically users don’t know when those are updated. There’s no regular pattern to it. It might update when someone’s house sells or when you pull a permit. And they may not record when the update happens. So you just don’t know how current that date is. With our data, not only can you get the currency, but we can also now look back in time because we know the dates other images were captured, and what's happened to this property over the course of time as the images were captured.
What is driving Arturo’s growth?
Clark: In a way, the COVID-19 pandemic has driven growth. We were around 25 people in March 2020. A year and a half later, we have more than 80. A lot of it is based on market adoption. Insurers were already looking before the pandemic to become more digital. So instead of me asking you a question on the phone as your agent, then typing it in and submitting that form, they asked themselves, “How do we make this digital?” But I think when COVID-19 hit, it really did two things to the industry. First, nobody wanted to have someone sent out to their home for an inspection to get an insurance policy, and insurance companies were concerned about the risk to their employees. So, how do you do those things without sending people there? Go digital. Second, people were spending tons of time in their homes, which created all sorts of demand in regards to things like remodeling, more potential wear and tear on properties as people spent more time in them, and the homeowner seeking a better insurance policy.
We were around 25 people in March 2020. A year and a half later, we have more than 80.’’
Tuttle: There are also just the larger insurtech trends going on. There are a lot of new companies coming up that are offering you a quote online in a matter of minutes without you ever talking to a person. Those companies need better, faster and more automated data sources to make that happen. We work with a number of them. That’s making the traditional insurance companies take notice and go, “Wait, maybe we want to look into that, too.”
Intellectual Diversity
What does Arturo’s approach to leadership look like?
Albrecht: I have an interesting perspective on this because I started out at Arturo as a junior-level developer. Now, I have seven or eight direct reports under me as I run a team. We’re definitely still a startup, so the organizational structure is flat, but it’s becoming more vertical as we get bigger. We have a great set of leaders who are good at empowering the people below them with more responsibilities. If you ask any of the individual contributors, people who are pushing commits into the system and actually building the tech, most people feel empowered on a daily basis, because we don’t tell them how to do their job. We just frame what we want to build.
Tuttle: I do try to embrace the leader-leader model myself (the leader-leader model forces you to push power and responsibility as low on the organizational hierarchy as possible). We have a ton of smart people. We’ve built more of an academic than a corporate culture, where everybody that we’ve hired brings a unique perspective. And we all come together and hear each other out to come up with better solutions. I think that’s really allowed us to propel forward at a much faster pace than your traditional, more competitive environments where departments are competing for resources.
One thing we always talk about is, it’s not ‘you’ or ‘me,’ it's ‘our’ or ‘we.’’’
Clark: My goal has always been to build a world-class team. It all comes down to agency. One thing we always talk about is, it’s not “you” or “me,” it's “our” or “we.” It’s our problem, we have this challenge. And so we try to keep that at the core, and then have the agency to go figure out an answer to the problem.
Now that Arturo is beginning to generate buzz, what’s next for the company?
Clark: We want to be the best at doing what we can do better than anyone else. Just because we’ve had success or raised more capital, that doesn’t mean we’re going to try a bunch of crazy stuff that’s outside of the core thing we do.
One of our current focus areas is addressing how we scale so that our business expands geographically. We have team members in Munich, and we support customers in Canada, Australia and New Zealand. We’ll probably support more customers in the EU and UK region by next year. But we will always be focused on doing the same thing and scaling that out to other parts of the world. Our vision is that we’ll be able to do it everywhere someday.
Tuttle: I’d love to pile onto that with data. One of our key differentiators is that we are a multi-source imagery company; we don’t focus on just one data provider to get our imagery. And there’s a lot of new things happening in that space, like awesome startups doing different things with new sensor phenomenology or new platforms using stratospheric balloons to collect that imagery instead of planes and satellites. As we grow, we want to make sure that we’re taking advantage of all the new tech coming to market so that we can continue to bring the most current capability to our users.
 (1).png)
