Roy Mill
Hosts James Benham & Rob Galbraith are joined by Roy Mill from Joshu. James, Rob & Roy discuss building better insurance products faster and lean business practices.
The full transcript
Transcribed automatically from the episode audio, so some names and terms may be misheard.
James BenhamOn this episode of the InsurTech Geek podcast, lean, insurance, and building better products faster with Roy Mill from Joshu. The InsurTech Geek Podcast powered by JBKnowledge is all about technology that's transforming and disrupting the insurance world. We'll be interviewing guests and doing deep dives into specific tech we see changing the industry. We're taking you on a journey through insurance tech, so enjoy the ride and geek out. And we are back with another episode with me, my illustrious cohost, the most interesting man in insurance,
Rob GalbraithRob Galbraith. How's it going, mister Rob? It's going well, James, and, we're finally getting some spring temperatures here in Texas. It's a beautiful day out here, and sure it's similar where you're at. Yeah. It's been great. I went and worked out at usual 05:30AM
James Benhamtime and outside of the park, and it was a balmy 66 degrees this morning at 05:30 in the morning. So I enjoyed that. I've been, freezing my biscuits off at about 33 degrees every morning, so I'm, I'm I'm happy to have warmer temperatures so I'm not starting the day out frozen. It's, it is a welcome relief. With us from a place that rarely has horribly bad temperatures from Menlo Park, California, our guest, Roy Mill. How's it going, Roy? It's going great. Thank you. How are you doing, James?
How are doing, Rob? Thanks for having me. Doing great. We're glad to have you on board. We're gonna we're gonna geek out with you. Before we geek out on, your tech, let's, let's just geek out on you for a second. We've had quite a few Israelis on the show. Mhmm. It's either popular for Israelis to be an InsurTech or popular in InsurTech to be Israeli. I can't figure out what it is. But, whatever it is, we end up with a lot of, Israeli InsurTechs. You worked like like many of your, compatriots worked for the Israeli Defense Force as a, in your case, as a programmer and product managers that obviously shaped some things around your your life.
And then you studied, politics, philosophy, and economics, nice, well rounded education, and then got a PhD in in the economics from Stanford. So what was the what was the the path, you know, starting obviously, you you had mandatory service in the Israeli Defense Force, but then, know, what led you to study what you studied and and your initial career? What was what was the story there? Yeah, awesome. Yeah, just a geek like other geeks, the IDF finds them and recruits them to do leaky stuff in the military.
Roy MillAnd I actually did like five years instead of three years, had to sign up for different things to make it more interesting and it was a really fascinating service. In the end, I decided I want to move away from tech and all the software and I'm going to go and study something about how to help the world be a better place through economics and politics and philosophy. And as a lot of other people from my unit and my co founder and Joe Shu Shimi who went to study film, we found our way back to tech.
You have the skill, you have this way to make money and something that is actually you're pretty passionate about. But before that, I did this quick five year detour of the PhD in economics that was actually, know, the plan was to become a professor of economics and in the end, kind of academia and I decided to part ways and find my way to product management, back to product management.
James BenhamYeah. I mean, that's goes how sometimes. I did a I did a five year stint as an adjunct teaching, and and enjoyed it. Right. My current volunteer gig is on the board of a university, Texas Southern University. So I I I have my involvement in academia, but, you know, the private markets are are what really turned my gears, and I really enjoy that a whole bunch. And I'm I'm sure you're it sounds like you're the same way. So when you when you actually got out of this, you know, not a slacking school.
I mean, went to econ from Stanford. You end up going to one of my favorite websites, ancestry.com.
Roy MillLike, was the connection to genealogy and ancestry? That's a great question. I went out and started looking at tech jobs. What was interesting about this specific role with Ancestry was that it needed someone who knows data, history, and software, something in between. And a big part of my thesis was actually working with historical data from ancestry.com. We borrowed data before they changed their terms of service to prevent some of the data borrowing methods that we applied. And I knew the census very well and matching records and a lot of that good stuff and bringing in some of the statistical background was really helpful.
We were basically tasked with helping folks find stories in all that structured data. A family tree and all the records are super structured, a lot of rich data that's there, and people a lot of times, it's harder for folks to tease out what stories are there. And when you use now with the amazing advancements in large language models, at the time, it wasn't as advanced, but a lot of narrative building and a lot of kind of highlighting of interesting stuff that is happening in your family history and you weren't aware of.
So that's kind of what brought me there. I did a few more roles there in other areas and the DNA data in the end. But yeah, that was a fun four years with Ancestry. Absolutely.
James BenhamWhat was your main takeaway from all that? That narrative data was at its infancy and was really going to explode onto the scene, or what'd you really pull out of that? Wow, I
Roy Millhaven't put enough thought into this. I think one thing that I remember is just this really meaningful product that you're building. That was a really awesome feeling, whether it's a story that you find, whether it's something new about people's heritage and their identity that was really important for them. I love insurance conferences, but family history conferences, when you go and talk to people and you get goosebumps from the story they tell you,
James Benhamthat's something I kind of miss, to be honest, but not to say anything about insurance. I sometimes get goosebumps from insurance as well, not it's different. It's a it's a rarer moment than than, you know, people's family history. Family history is very personal. Of course, it's gonna it's gonna naturally be very interesting. It is. So you you went to be VP of product at At Bay, which we've we've discussed and talked about in, you know, news and things that At is doing. What led you to At Bay?
Was it just a random job posting or did something about insurance really, really
Roy Millpique your interest? Yeah, was the contrary, actually. I was not thinking about insurance. I was like all the other people who are not in insurance who think insurance is boring. But my college roommate from back in Jerusalem, she was the head of people for At Bay and she asked if I wanted to lead product management for the startup and I said, let's talk and found out it's actually super interesting, kind of combined my statistical AECOM background for the insurance side and the cyber side from some of the things that I've done with the IDF.
So I was very excited to join the team and build the product, launch one time, two times until we got that traction, that zero to one,
James Benhamall the way to the point where we decided to kind of leave and start Joshu. And what was the what was the the founding story of Joshu? There's everybody and I know I I've got mine from my different product companies. And, what was the what was the original thesis? What was it started for? Yeah. We
Roy Millworking on Ad Bay's cyber insurance product and it's a non admitted ethnic losses at the time. You're as fast as you want to be. You're not waiting for the state to approve your product. If you have an idea, this is the questions I want to ask about the insurance, this is my underwriting model, this is some more data that I want to collect, you're as fast as your team can bring those products to market. And we had a lot of back and forth between the insurance team and the tech team.
So underwriters, a lot of times, they want as much information as possible. And from a UX perspective, when you're building a portal, want as few as possible. And this back and forth, we launched twice. The first time was a very long application form for larger companies, then changing to small and medium businesses and much shorter applications. Each time, it took a long time to do that. And we felt at some point that we're kind of getting in the way with us building everything custom and bespoke for this use case.
And we said it must be a bigger problem than just AdBay. So how can we give our insurance colleagues at AdBay and beyond a tour they can be more independent, where they can actually take their idea of an insurance product and configure it with that being technical on the software side and be able to launch it with an end to end kind of digital distribution, underwriters portals or not portals or desktops, workbenches, and have the whole thing kind of set up for them. And is it and so what lines did you start with cyber?
No, we're pretty agnostic. We intentionally wanted to build it in a way that would fit different use cases. We had a few kind of early design partners, some of them in kind of techy, know, others in property, actually solar projects and things in that nature. And that combination allowed us to build the platform to be kind of more agnostic. Obviously, as you get to different product lines have different requirements and some of them are better fit and some of them will be able to kind of cater even better as we build on more capabilities into the platform.
But the idea is, you know, as long as the apply quote buying flow is what you're looking for, we have, you know, a platform that's ready for you to build your product and test it out with the market. Great.
Rob GalbraithGreat. Rob? Yeah, Rob. I'm really fascinated, with kinda what you're doing, Josh, and and your background. And I, at one point considered, going down the the PhD path and and never started, applied to some programs, but never ended up going down that path. So, yeah, I just kind of curious, we'll have to talk offline sometime about, kind of, you know, what what was it about academia that kind of led you down, this path instead as an alternative? But, I just would love to kind of, maybe share a little bit more about Joshua and what is a week in your life kind of look like?
Like, who are you working with? Are you working with, you know, tech folks? You reaching out for, you know, partner companies? Do product testing. I'm sure you have a lot of balls in the air. Maybe just give our listeners and viewers a sense for what a typical week is like for you, Josh. Yeah. And a typical week is now different than a typical week when we started. The nice thing about early stage is you start doing everything.
Roy MillKnow, Shimi, my co founder and I even coded together when we were just prototyping back in the day because it was just the two of us in selling and doing everything. We're very lucky to have been able to kind of bring in a lot of great people definitely on the technical side, on the sales side, solution consulting. So currently, actually the last couple of weeks, most of my week is spent on fundraising. We're reaching out, talking to different investors on our next upcoming round. And you combine that with basically blocking, tackling, helping people with the sales process, getting more customers on, product discussions with the product team.
It's kind of all around, but the nice thing is I can focus more. Recently, I've been able to focus more on how to get the company to scale and to get the funding and then the teams and the vision. So talk to me about this because
James Benhamare you doing full policy administration, policy issuance, certificate issuance,
Roy Millunderwriting workbench, billing? Is it the whole suite? Yes. Billing, the last one right now, is provided through a third party integration. And a lot of folks have different they already have a relationship with one provider or the other, and we're happy to make integration. But the policy admin system, we do get this question a lot of times, are you a policy admin system? And there's some overlap with this. With the new insurers that we work with, a lot of times they would only need us and they don't need the full policy admins, the full guidewire or a full ductwork for that.
It's kind of a small scale. We power those transactions, apply, quote, bind, issue, endorse, cancel, and they're good to go. With the more established players, they already have a policy admin system. The way that Joshu works is more of a product lab for their product team so that before they go and implement a long term project with their policy admin system, they're able test out the market, test the concepts, see what works, what doesn't and kind of scale from them. That kind of goes back to the lean insurance and lean carrier approach.
Try with a small investment, see how much you can de risk in that new product idea, product launch, you're planning and see what you can get from the market to get a better product out there. So in other words,
James Benhamyou'll actually, they'll essentially be running two policy systems until they decide to bring that line of business into their core policy system. For the established players, it doesn't have Again, we can work with just a prototyping tool that not
Roy Millactually sell real policies. We're happy to The thing is the big policy admin systems are work great for the standard lines. You get into specialty and get into some of those less standard lines and things become really long and people are not getting their right love and attention and it's not a great fit sometimes with that big platform. So they can run everything on Joshu and we will have an integration through the policy admin system for accounting and billing and all the other reporting with the states, things that happen anyway on the back end.
But at least on the front end, and the idea is not to have two systems that you have to configure twice, but one system you can configure really quickly based on what you get from the market, based on how you want design and work on that trade off between underwriting needs and distribution needs, and then send off the data back to the mainland when you have it. So initially, it can work with file drops, and then we can build an integration as well to make sure that it can work, kind of, can scale.
Yeah. And talked to some folks in
James Benhammajor insurance companies. They even have rules that have until a line of business gets to, you know, x number of tens of millions of dollars or $10,000,000 or $50,000,000 write all of it in Word and Excel. Their company has a policy largely because of decades of scar tissue building up at those insurance companies where they invest millions of dollars in technology spend ramping a product lineup. It doesn't work out or it's not a big enough market or they decide to pull out later and they just blew this massive amount of money on technology.
And so we're seeing an interesting evolution where companies are saying, Hey, you know, actually, can manually underwrite this And write the policies until it gets to a certain size, at which point we can change the way that
Roy Millwe're writing this and do it with software. Are you seeing that a good bit in your client base where you're saying, hey, we're going to prototype this product in Excel and Word? I do. And I think I see a lot of people sometimes say, well, we don't want we don't like the concept of a portal because everything is very manual. Some underwriting is like, after twenty years in the industry, I feel like this should be $12,000 of a premium or $25,000 There's not even a rating model per se.
It's someone just sitting there and having their expertise and building a great book of business, but it's very manual. And they end up spending a ton of time on very menial tasks instead of actually underwriting, and we tell them, hey, you can just you don't have to have the portals. You can just use this as a way to streamline your underwriting. And you hit the nail on the head by saying, home and auto, definitely a lot of premiums, small premiums, very automated, very standard. You get into commercial lines and different businesses, the risk profile of a nursing home is nothing like the risk profile of a law firm, right?
And the types of products that they need are really different. So you get into this matrix of what types of businesses there are and what types of coverages, and suddenly the premium, if you aggregate the whole kind of matrix, it's a lot of premium, but in each cell, it's not a lot of premium. If you start implementing those heavier policy admin systems on those, small cells of business, it's just not worth it. So that's where we come in and can be way leaner and kind of provide a solution there.
Awesome. Rob? Yeah, Roy, I want to talk a little more on this concept of kind of lean insurance,
Rob Galbraithright? Because this is a new concept that I haven't heard about. Obviously, I think many of us know about kind of lean startups. You know, prior to recording, we were kind of chatting, and you talked about some of your favorite innovations and insurance aren't necessarily right use of AI or GPS, whatever, but like the product innovation. And, obviously, Josh uses, like, a lot of that in a way that there there's ideas in folks' head, but, you know, we've just touched on all the limitations. So maybe you can just expand a little bit more when we talk about lean insurance.
Like, what do you mean? What does that look like? Help our audience kind of understand like, hey, what are the types of positions or when folks are coming to you to see if you're a solution, what are they trying to accomplish? Great point. There
Roy Millare a few ways to go about it. One first is to kind of talk about the problem. As James and you mentioned, it's very long. It takes a long time to launch a product. And people are always you hear other core systems say, we're the fastest. No, we're the fastest. And people measure time to launch. And people don't realize there's after the launch a lot of times you launch, but there's nothing there. You open a bar, nobody comes in. You open their pizza place. Well, I got the location wrong.
I got the pricing wrong. You need to iterate on this concept. And the concept of lean startup and kind of agile software development have evolved in the software world where they had a similar back when you had to burn CD ROMs and diskettes and send it to people or to stores, it's the same process. People have the same process because you have to go through months of development to close a version and testing and then send it off. When the web came along, it made software so much leaner because you could change the web service, the software that runs for browsers in a second, and everyone gets the latest version whenever they come on board.
So that kind of shifted people's mindset and said, hey, we don't have to cross all the Ts and dot all the Is and make sure everything is there. We can start small instead of having everything figured out. Let's think about what is the core, the hypothesis that we're trying to prove, right? If I'm launching a new management liability product and I have this concept of how am I going to distribute it, why is it going to work through retail brokers or why is it going to work direct if I'm going to sell it to startups or if I'm going to sell it to a specific segment?
And you're thinking about, okay, how do I get this product up? I have an admitted product or I need another specialized product for that, and you have this concept of, okay, I want to build this and I want to launch. And what we're seeing is in a lot of cases, the concept is great, the dream is great, you launch, and then you realize, actually, it's not exactly what we need. The market isn't responding as well as we thought. You don't want to spend $3,000,000 on something that then creates zero premiums, right?
And that's just a waste of time. So the question is, how can you invest the least in finding out if your hypothesis that you're going to sell management liability products to this segment is actually working? If there are ways to build a simple product and then iterate on it, see that is the right distribution. No, that's not the right distribution model. Actually, we don't need 20 questions in the applications. We can do away with like 15. And maybe there's another data source that we can integrate and enrich our underwriting and get a better read on those risks.
And all that together, the question is how fast can you get there and do you have the tools to experiment through this? Or is it something that requires a lot of committees, intake process to get IT resources, you get them for three months, they go and they launch, and then you're on your own because they're moved on to the next kind of project. So for us, this independence of the insurance product owner is super important. We want them to own this and to drive this and to do as much as they can without our help or the IT's help.
Obviously, lot of them are busy and they say, hey, I'm also underwriting. The product build out is something that I do on the side, like 20% of my time. Can you build it for us? We're happy to help with that. But again, the concept is even if we do the setup, you'll be able to make changes or we're going to do it together so that you have we really want them to have the tools and us to design the tool in a way that really matches what they need in order to think about innovation and
James Benhamgetting that next iteration out there. Very good. Let's wrap up with this one question. It can lead to a bigger conversation. Low code, no code or configurable software? Can you explain the question? I told you it's big. So I'm a I've I've been writing code for a long time. To 1991, I started with GW Basic, then I moved on to Fortran, Pascal, Assembly, c, c plus plus, then jumped into, ASP, asp.net, c sharp, MVC, SQL 97, 2,000. Like, right. Okay. So I I did a lot of similar progression.
A lot of other people did. Mhmm. The the the longer I've been writing code, the less code we write. I mean, just in general, there's a lot more intrinsic functions. Mhmm. A lot of stuff's built in now. Right. We don't write. Mhmm. We write probably 70 or 80% less code than we used to because we don't have to. Right. We we call a function. We call it we call a service in Azure. I mean, we it it thing things happen for us now. Mhmm. The the hot buzz phrase in policy administration, underwriting workbenches, etcetera, and then just in in product is low code, no code.
To me, this concept is not new. Right. Guidewire invented a programming language called Gosu, like, a long time ago, and you had to learn Gosu, and then you coded your product. It just took a while. It still takes a while. Mhmm. You know, Duck isn't that different. Majesco isn't I mean so you you either have a configurable system where you just configure variables infinitely, or you do something like some other companies, I'll I'll they'll remain unnamed, that that basically did a JavaScript variant, and they built a low code platform.
What do you think the answer is to
Roy Milleffectively service and spin up products? Thank you for clarifying. And I think that's where you were going, but I wanted to make sure. I think basically what you're referring to is this there's this trade off between how simple your platform is and how powerful it is. You want to make it simpler, you're dumbing it down, or you want to make it more powerful. You're like, Okay, now that's how do I do this? And back to the econ, right? We all want it to be economists or are economists.
And we're saying, think of it as like a frontier between remember when you're trading what was it? Roses and guns or something? Guns and butter. Guns and roses. Guns and butter, right? My guns and roses, They're Right. Better And the question is and usually what brings that frontier up into the right is tech advancement. And to me, it's less about code or no code. Sometimes there are some platforms, the more generic kind of no code app builders, right? It's drag and drop, right? So you didn't write a single line of code, but you need to really know how to drag and drop those elements so that it makes sense and they don't break.
In the end, it's what is that learning curve that you need to walk and how steep it is in order to deliver value and how quickly can you deliver value. To us, that's the biggest kind of I think there's a tipping point because if you still need to be It's low code, but still you need code and a lot of people are scared of it, right? So if you can get to a point where you don't even need the low code and it's really You can measure your platform and how well it does on self-service in terms of what percentage of product owners, insurance product owners, can set up simple products, middle complexity products, and high complexity products.
And I think that's for us the product vision and the North Star outside of providing that kind of basic transaction support that can policy admin system like is that self-service promise and vision so that people that are not technical from a code perspective, they're technical maybe from an Excel or insurance product perspective and language forms and things like that, they'll be able to use the system. If they want to dedicate the time, it's very easy for them. And they can get from a simple product to their own product that is more complex than that, and we provide the tools to get there.
Gotcha. Good answer. I think it's a really tough problem to solve. I think we have amazing people to solve it with, and I usually invite folks to challenge us and watch a demo, see if it looks simple enough for them, and then they're like, oh, this looks simple, but it's complex. Our radar is much more complex, and I tell them, challenge us. Send over your rating model, your product and we'll see if we can build it. Sometimes we say, listen, this is 50 states admitted products, commercial auto, probably not a good fit for us right now, but your dentist professional liability that you thought is super complex, we here it's done end to end flow in less than twenty four hours.
So it really depends on the use case. In future, we want to cover as many cases as we can in terms of our support.
Rob GalbraithObviously, the complexity levels Yeah. Can vary. Excellent. Rob, you wanna wrap us up? Yeah. Roy, just quickly, you know, where can folks find out more and then what's kind of on your roadmap for the next twelve months? Yeah. So find out more on our website, Joshu. It's joshuins,likeinsurance,.com,
Roy Milland that's our website. Social media, LinkedIn, we're pretty active there. That's a great way to kind of follow us and see what we're doing. So for us, the of the roadmap from a product perspective is, again, wrap up some of the policy admin system stuff on the edges. The core is there. There are some things on the edges that we need to complete and keep simplifying onboarding, make it easier from a video instructional, make the product initially simpler to work with so that we deliver on that promise.
And then once we have these two in place, I think there's a lot to be said about insights and how you use that data to optimize. The way to iterate and say from no adoption to adoption, it's pretty clear if things don't work, nobody submits any business to you. But sometimes the optimization, yes, you're getting business, but people stop midway and this question is really complicated. And you can measure how long people take to answer this question. And combined with your underwriting model, we think this is what you can do better on this product.
Those kind of next level
James Benhaminsight, I think, are really great to optimize products and make them better. That's great. Well, thank you for coming on and talking about this with us. Really appreciate it. Always good to have a chat about simplicity and the trade off, and this is like the the fundamental existential
Roy Millargument in software. The trade off between the trade off between simplicity and power. It's It is. It's and any any software engineer will say they're they have this choice between using a framework and bending it to what I need or starting from scratch and just
James Benhamthrowing it away. So we're all over it. Yeah. Nicely put. It's a lot to take in. It's a lot to discuss, but I appreciate you being on. Discuss it, and thank you for the excellent conversation. Appreciate it. Awesome. Thanks, James. Thanks, Rob. I appreciate it. It was fun geeking out with you. Yes. And, thank you to all of our listeners for tuning in today to geek out. Prior interview with Roy Mill from Joshu. This has been the InsurTech Geek Podcast powered by Jibby Knowledge. It's jibbyknowledge.com, all about technology that is transforming and disrupting the insurance world.
I've been your host, James Benham, jamesbenham.com with cohost Rob Galbraith, endofinsurance.com. Thanks to Jim Greenlee, our podcast producer, and Cara Daltonaro, our creative producer. And thank you for joining us today. We're taking you on a journey through insurance tech. So enjoy the ride and geek out.

