Graham Elliott describes Azur as a "Managing Digital Agency" (MDA). He founded the business in 2015, providing High Net Worth insurance for clients of UK brokers, in partnership with AIG.
Matthew Grant and Graham talk about the challenges of becoming a full digital insurer and other topics such as offering quotes based on only 5 questions, sources of data and the role of brokers.
You can learn more about Azur at https://www.azuruw.com/
Transcript for this podcast
00:00 Graham Elliott: If you look at the activity of a lot of data scientists and big companies, I would argue that 80% of their time is spent cleansing and finding and trying to mess around with the data and 20% of the time is spent actually analysing and drawing conclusions from it.
00:24 Matthew Grant: Hello and welcome to the Instech London podcast, this is Matthew Grant, one of the partners at Instech London. Now, I enjoy doing all our podcasts, but this particular episode was fascinating. Graham Elliot is the founder and CEO of Azur Underwriting that he set up in 2016. Now we're often looking for examples where the full stack digital insurance might come from at Instech London. And in talking to Graham, it's clear that he had been able to very successfully combine both technology and a deep experience, not just insurance, but in other areas of business to provide a MGA that is adding value both to their clients, but also all of their partners in the full value chain. I think you'll enjoy this a lot. Graham, thanks for joining us today.
01:14 GE: It's my pleasure.
01:15 MG: Now, you've had an interesting career, you started off with a degree in English from Oxford and then you've been in marketing, the oil business, you've been in banking and you've ended up in insurance. So is the conclusion of that for anybody out there wondering what career to take they should really just start in insurance 'cause that's basically the best place to build your career.
01:38 GE: So basically what you have got to do is choose choose a career where there's a lot of energy and where the economics are good. Banking is a phenomenal industry for a long time, but post the crash and post the melt down, it really was a less interesting place to be, and the particular job I was in was basically stock broking, which was unbundled by the regulations so it became even less interesting. Insurance is phenomenal because it's the mutualization of risk, which means it's spreading it and fragmenting it through society. And as a result of that, the players are well fragmented, and there's lots of little players in there and it seems to me to be a natural entrepreneur's landscape. It really is an amazing playground for anybody with an entrepreneurial bent.
02:23 MG: And just to that last point of being the entrepreneur's playground. You've been fairly vocal, maybe more in private than in public about some of the challenges you've seen in the insurance industry. So what is it that drove you to start up Azur, having been a broking role before and just given some of the challenges that you see out there generally in the insurance space?
02:44 GE: Well, I was running an MGA in a broker before, it was great fun, it was really enjoyable. But I could see the problem was that we didn't have any control over our technology and that meant we didn't really have control over our data. And insurance is a $4.3 trillion industry with a massive data problem. And to be a 21st century business, you've got to solve that problem. And so the only way was to start from scratch and try and take advantage of modern technology and set up your own business and have control over your own technology and that meant becoming a technology company as well. I couldn't see doing that from a legacy player. It's very hard to retrofit that type of culture.
03:31 MG: So many insurance companies try to claim that they are, or some of them try to claim that they are, actually data companies or technology companies in insurance. The reality is that, they're actually insurance companies. But given you started off three years ago, and you got technology platform, and you got an MGA, which of those do you tend to put forward when you're defining what you do, are you an underwriter or are you a technology platform?
03:56 GE: We're a technology-enabled underwriting business is the way I would look at us. And it's very difficult for people to get their heads around, "What are you, you're a technology company or an insurance company?" I don't think that's a 21st century question. I think in the 21st century, they're going to be companies that own their technology, control it and know what they're doing, and right the way through the organization as a tech-enabled business in the cloud, and I think the other companies just won't be around in the end of the 21st century, so I think it's a must do. And it was the only way I knew of answering that problem.
04:29 MG: Yes, essentially, the data and the analytics is a basic requirement for entry into the business. You need to have that if you're going to be a successful underwriter and then you make the money from the underwritings.
04:39 GE: Yes, I think that's right. I mean, if you look at the activity of a lot of data scientists in big companies, I would argue that 80% of their time is spent cleansing and finding and trying to mess around with the data. And 20% of the time is spent actually analysing and drawing conclusions from it. And that's because the core systems that those businesses sit on weren't built for the age of data, they were built for counting money and it's not the same thing.
05:06 MG: And you talk about being a Managing Digital Agency or an MDA as well as a MGA. Can you just explain what the MDA?
05:15 GE: Yeah, I thought we had to... It's a completely cheeky piece of land grab. We thought we had to invent a category because we didn't really fit into anybody's category and when we went around originally trying to fund the business, everybody would go, "What are you? Tech or insurance, you've got to be one or the other." And I couldn't see why I had to be. So in the end, we thought, "Well okay, we'll call ourselves what we are." And it's aspirational, it's a journey for us, we inherited a legacy portfolio and we are in the process of digitizing that, but the new products that we're building and have built are all fully digital.
05:53 MG: So you are building the elusive full-stack digital insurance company, from the start.
06:00 GE: 100%, end-to-end. And it starts before the quote. It starts at the engagement level, when you're dealing with, "How do you engage people." If you're in a B2B context, how you are you engaging your audience of brokers that you're selling through. And we do that through Broker IQ and all of the data there is on the same platform as all the data about the brokers themselves and the risks and all the financial data at the back-end is there as well. So it's all joined up.
06:28 MG: And I'll come back to that in a minute because there's a lot of challenges around that... You're certainly overcoming some of them. Quite interested how you deal with some of the other issues of data that's beyond your control. But just with regards to the book you inherited and the partnering with AIG, a couple of questions on that. So first of all, for a company like AIG that's got it's choices of who it can get involved with and do a lot of things internally. What was it about Azur that ultimately AIG decided to support you and give you access to their book?
06:58 GE: I think what they liked was that: A, I got experience running that type of business before. So I had got domain knowledge. And they liked that I got a background, somewhat of a background in technology. I chair a technology company. And I think they liked the vision, that instead of writing premium, which the carrier had appetite for, your MDA or MGA actually writes client-facing stuff. So we're a great transformer from vertical risk appetite into horizontal client needs. And I sold them on the idea that that was what was going to drive the intermediary of the future, doing this stuff, that you could build products for clients that suited them and that were digitally enabled to be very personalized.
07:54 GE: I think technology, from the printing press onwards to the PC, really delivered one-to-many increasingly quickly. And what cloud does is allows you to deliver one-to-one at scale. And the metaphor for that is, think of your Facebook user. There are two-point-whatever billion users of Facebook. There is one version of the code and everybody's got my Facebook page. It's configured for you, so everybody's page is personalized to them. And that's the sort of experience I think is going to have to be the expected experience of the future, both B2B and B2C. So the millennials coming into a legacy business are going to look at the platform running on green-screen IBMs and go, "This doesn't do it for me. I don't want to join a company like this."
08:39 MG: We'll come back to legacy in a moment. Just on the AIG book, so this is the high-net-worth book that you brought over from AIG.
08:46 GE: Yes.
08:47 MG: So when you talk about vertical versus horizontal, just explain a little bit more what that means in practice.
08:52 GE: So for 300 years, approximately, insurance with written for capital, and the metaphor for that is the Lloyd's building, it's a temple of capital, it's a great place, phenomenal expertise. And the brokers would queue up and they would place risks, but the risks were what the capital had appetite for. And so no one capital provider can write all of the appetite for all of the clients that it would seek to serve. And so we buy insurance in a fragmented way because of that, and it doesn't seem to me to make any sense. And the MGA is the perfect transformer vehicle to take pools. It's agnostic as to capital, so it can take pools of capital with different appetites and string those together for a client solution. And that seemed to me to be logical and quite exciting. And if you look at the data models of some of the legacy enterprise software providers, they recognize an insured object in the data model, they recognized an insured person. But they don't have a client in there, necessarily. Well, that tells you everything about it. It was built for a different era. And now we're in an era where the customer's king and distribution has won out over capital. And we're moving into product now, and services. And you can't do those things unless you're really personalizing it. You can't do that unless you've got the right vehicle.
10:08 MG: So it feels like you're starting to answer the question that anybody about insurance at some point asks themselves, which is, "I'm I insured for this particular event? And how do I figure out from all my different policies if I'm covered?" In the moment, you do, I think, personal lines, you do collections, you do motor, so it feels like you're building out a portfolio. So at some point, potentially, one of your clients knows they're going to get coverage across the full range, 'cause you're making it solution-centric rather than individual products. Is that how...
10:37 GE: Yeah. Yeah, I'd like to be able to do multi-niche insurance that caters for client needs. So one by one, we'll knock off the big things. At some point we'll do yacht, and then we might look at at equine. And so you're building out this portfolio of products and the client can pick and choose what they want to have. And I believe that people will be happy to have that convenience.
11:00 MG: I'm quite sure the Grant family qualify for high-net-worth in Azur but certainly at some point we'd be delighted to have an insurance policy where we were confident we were covered for all losses and not have to figure out what we were covered for, what we weren't covered for, and what we were paying three times for.
11:13 GE: Well, contact one of our brokers and I'm sure we'll look after you.
11:17 MG: Fantastic. So let's get back to this legacy one. This is an issue that a lot of people use as a barrier to growth. You've got some ideas to think about how to extract data from legacy in an efficient way without necessarily having to rebuild the systems. Is that right, in terms of how you access that?
11:36 GE: So the first thing is, it's a lot easier to build new products on new systems because often getting off of legacy is not actually a technology question as much as a reimagining the way that the thing is underwritten. And so we've launched a new product in the UK, but our the real passive goal is to digitize legacy. If you've got multiple endpoints because you don't have them fixed on a system of record, then it's very difficult to build any toolkits that transforms from one system to another because you've got both ends moving around. If you fix one end and say, "I absolutely know that's where my endpoint is," then you can genericize your toolkit and begin to build tools that can do the ETL part of this. And therefore you can have digitalization of portfolios at a lower risk. And that is... The benefit doesn't just accrue to the customer and the broker, the end journey, the end experience. The benefit accrues to the insurer because you're taking away a ton of processes, we-keying and mantronics, and the orgy of terrible manual labour that goes into all this stuff. And you're stripping costs out, which is what everybody wants to do. And in the end, that leads to lower premiums for the same economic result for the insurer, which leads to a better result for the end insurance. So if you can get after this problem, it's a big problem.
13:03 MG: There were a couple of interesting acronyms and words in there. So ETL, extract, transform, load.
13:09 GE: Yes, absolutely.
13:10 MG: Mantronics, did I hear that?
13:11 GE: Yes. It's our word for, it's the Wizard of Oz, basically. It looks like everything's... And there's somebody behind the scenes pulling levers to make it all happen. And there are lots of insurance businesses where stuff is put into one system and then somebody in a centre somewhere is taking it from one system and keying it into another, and that's madness.
13:34 MG: Well, mantronics, you heard it here first. Good. Just one thing you just mentioned in passing, I want to pick up on. You mentioned the broker there. I believe, at the moment, all of your submissions go through a broker rather than going direct. And what's your view on the future of the broker in insurance?
13:54 GE: Well, I'm a firm believer in the value of risk transfer advice and the need for it. Beyond the most simple risks, you really want to understand what you're getting into. The broker has an absolute role in this. And we're not trying to disrupt the broker because I think that's not a particularly logical place to start as a small company. "We're going to disrupt Aon." Well good luck with that. Of course we're not. They're phenomenal companies, they're very well-run. We view the broker as being very necessary. Do I think the broker's role is going to change? Well, yeah. There's a lot of places where it has changed. I was a stock broker. And that un-bundled. They decoupled advice from the execution of that advice. So the fund would only pay for the research itself, and the underlying company would not pay, the underlying fund holder would not have to pay for the research. And I think it's hand in IFAs where you got RDR. And there is a chance that they decouple advice from the execution of the policy. And it's not a percentage anymore, that doesn't appear to be any time soon. The role of the broker's absolutely there. They've absolutely embedded the importance that they've got in the whole risk transfer process. And I'm a firm believer in it. We're going to work with them, not against them.
15:16 MG: Good. So you can still walk down Lion Street safely.
15:18 GE: Well, obviously. Not wearing a shirt. But, yes. [chuckle]
15:22 MG: Good. But I know, also just to clarify that is part of a factor of the high net worth business. So you've got complicated needs. It's not like the areas where, in the UK in particular, you've got this retail brokers going direct just 'cause they are fairly homogenous transactions in place.
15:34 GE: Yes.
15:36 MG: You started working with Logical Glue, another civil organization, it's sort of been around a few years now. And also, I don't know if this is what they're doing or is happening separately but interested to learn a bit about how do you achieve your pricing where, I believe, if the broker puts in five attributes, you can give a response back in 90 seconds.
15:56 GE: Logical Glue is slightly different from, our work with Logical Glue is slightly different from the data enrichment that we're doing but they belong with each other. Behind all of this is my conviction that we're going to go to a world of augmented underwriting. And augmented underwriting is we're great at things as human beings, and you still need human beings, but you could be even better if you had a bit of help from the machine. And in two ways. Firstly, if you enrich the data, you have the chance of delivering the asymmetric user experience that's so necessary for a 21st century product. If you've got five questions on the happy path for the broker and the end insured then that's a pretty painless experience to get a quote. And you can do it in 90 seconds, versus 40 questions, and it's taking you 25 minutes.
16:46 GE: At the back end with data enrichment, the carrier in our case AIG, gets 66 rating variables. So by using data enrichment that way, you give a great UX both ends. You satisfy both needs. Logical Glue is the bit where you take that augmented data and then you start to have the machine watch and help guide you on which risks you want and which risks are going to be difficult for you. It's not you're cherry picking, you're just guiding yourself and trying to help yourself avoid some real pitfalls. And we think there's some really interesting stuff that you can do, and I would hesitate to call it AI. I think it's more explainable machine learning, and even then it's not actually active machine learning, where the machine is correcting itself every five seconds.
17:40 GE: But you have a model and you train the model and then you run the data, live data through versus historical data, and you work out whether you're likely to win a risk and whether it's likely to claim or not. And it doesn't stop you, the underwriter, from making the decision, but it just helps you. And there's a lot of people on Wall Street who use quant underlays to help them select stocks, because there's just more data out there than you can process as a human being. So you have a strong instinct and the machines are saying to you, "You know what? You're right on this one, I agree with you." Or, "You want to be careful." And then it'll tell you why. And you might still go ahead and underwrite it. And that is applicable not just to a high volume class of business. I think that augmented underwriting is going to be an absolutely common thing in 10 years time. Everybody will be using it.
18:27 MG: Yeah. And just the theme seems to be moving on from the underwriter no longer has a role to the underwriter can do more efficiently and more accurately, but it still needs to be there for those referrals or the tricky, tricky adjudicators. So does that mean that you have underwriters here for certain situations that might get flagged up by the tool in the background where you actually have to have some human interventions and not everything is going to get...
18:51 GE: Yeah, yeah. No. You never... Because we don't fit into buckets easily, our goal on the higher volume products are to have 80% go through on a happy path. We're not there yet, but 80% on the happy path and 20% have to be touched. The 20% that have to be touched, what you want to look at is, "Do I want to discount this risk anymore, or do I not?" And that's really helpful. Because otherwise, you've got no basis for knowing other than what the broker's telling you. And of course you've got to trade and you've got to be there, but you know if you're going to have a quote to buy in ratio of 20%, make sure you got the right 20%. And the distribution of claims through a claim series is not linear by value, so the bottom 5% of claims will be 40% your value. And so if you can avoid one really bad claim, because the machine says "Hold on a minute, do you know this is quite difficult here?" And if you can avoid one of those, you can make a big difference to your book. And you haven't screwed up your ability to trade with brokers. You just helped yourself select the risk a bit better.
19:57 MG: Yeah. And that makes complete sense. So with the pre-fill, I can see in the next year or two there's going to be a race to see who come up with the most, I don't know if it'll be the fastest 'cause at some point that doesn't become relevant. But yeah, then the most effective way of doing those pre-fills or rating on the basis of five questions. Can you talk a bit about what you're using as a data in the background that you augment?
20:24 GE: I think data is going to commoditize more and more, and the data companies will merge and you'll get bigger and bigger data sets. One of the problems for a niche business, is you don't have breadth, horizontal breadth of data. So you don't have big data horizontally, and so therefore what you want to get is big data vertically on each risk. We have access to 27 million mortgage surveys, so we have property-specific data. We're looking at floods, subs, sanctions, everything else. And then we've got a lot of modelling going on in the background around confidence scores on how many bedrooms it's got.
21:02 GE: How many... What the rebuild value is, what the square meterage is. And we're always looking for ways to improve it. But when we back-tested the manually entered, manually given data that we got, we found out that quite a lot of it was inaccurate, a decent percentage was not right. The machine is just less wrong than the human. We do place a hard burden on the human... On the machine than we do on the human. So everybody goes, "Well, you've moved from 60% accuracy to 80%, that's terrible, what about the 20%?" Well, what about the 40% that the humans are getting wrong? So we got this journey and the data is not out of the box. And if you try and use it out of the box it's not plug and play, so you have to spend a long time testing it. And you have to spend... We've spent a lot of time on rebuild values and we've gone street by street, sometimes in London. So not even post code, not even second half of post code, street level rebuild values. So it becomes highly proprietary to the company, and that's pretty cool.
22:04 MG: Yeah well, you picked up on a hot topic of mine just now actually, which is rebuild values. Partly because in the US, if somebody under-values their building, they get penalized when it comes to the insurance. UK I don't think is doing that or is not doing as much. But now people tend to be advised to use the ABI Rick score where... And just having gone through a renewal myself and looked at the rebuild cost and knowing a little bit about what these things really cost. Seemed like is it way off in terms of it's estimates. So, yeah really intrigued about that. Can you also talk a little bit about how your... You take every street by street, what does that actually mean in practice?
22:39 GE: Well, we have our own surveyor here who's an expert on this stuff and he surveys high net worth houses. So, we look at... We're just very concerned that if you have a property in Battersea and the same square meterage in Knightsbridge, the rebuild value is going to be different. And it just is, and what is that number? And I don't think that the available official data is fit for purpose on that or it's a starting point, but it doesn't get you there. So we've had to go through manually and work it out. Once you've done it, it's pretty cool. And then the other data source we get to help us is we were worried about if you had a mortgage survey 10 years ago, but somebody's rebuilt their home or done a big extension. How do you pick that up? So we're taking planning data as well. And then what we started to do is contact the brokers and say, "Did you know the following properties of yours have had planning permission granted? And by the way, here's our course of construction insurance, if you should need it." So you can start to turn this data into a demand generation motion rather than just as a sort of passive source of underwriting, that's quite interesting.
23:46 MG: Yeah. And also actually, what's interesting as relatively new company that you feel you can justify the cost... 'Cause it seems like there's not a short cut to this. In a sense, you need to be as good as one of the big general insurers out there albeit I guess, you're looking at a slightly smaller section of the market, if you're focusing on high net worth. But it's interesting that you've been able to presumably do this in a commercial way, so you can get the data that you need and still you leave enough profit on the underwriting premium with the cost of the data coming in.
24:19 GE: Yeah, we're not subject to quarterly earnings report as a driver. And so we do get to be able to invest. But we wouldn't do it if it wasn't economically sensible. And the other reason why we can do it is because we've got the core system and all the data in one place. Sometimes the problem is going to be, you're going to get overloaded with big data coming in and your core system can't cope with the big data that's coming in, so you can't actually use the data coming in and you haven't got the underwriting to be able to cope with that. We're able to innovate carefully. We're not chucking money at it, but we're really thinking about what we do. And we're trying to keep ahead of all the time of where we think the world's going to go to and try to invest it for that because that's the only sensible thing you can do.
25:06 MG: And just one more theme around the data one. Water leakage seems to be becoming more and more of a topic, it's been out there for awhile. You're doing some work around smart homes, you've got a partnership I think with Grow in terms of their water detection device. I think it's also shut-off valve, is that right?
25:27 GE: Yeah, I am. It's interesting Matt, because... And it is just a pilot to work out whether we can help it. The trouble is, when you think about this through the lens of the end customer, and the user experience, which is our number one obsession, it becomes a bit difficult this because it's a bit like capital rights, what it's got appetite for. We'd all love every home to have a great water leak detector and a shut-off valve because then it would lower water damage. The trouble is getting all of that fitted is actually not trivial. You've got to have somebody at home, you need a plumber, then the plumber arrives and they realize they need a plug to be installed near where the shut off valve's got to go and there isn't one, so you needed an electrician. Then you need to have your WiFi, and then you realize the WiFi signal doesn't go to the scullery or wherever it is that you have to go to. And so you then need a WiFi expert comes in and not everything works first time. And by the time you got through that, and then the water leak detector goes off on your wife's phone at 3 o'clock in the morning, you have a failed marriage on your hands. So I...
26:33 GE: We're going to try all the stuff. I think there's a lot of hype around it and we're just trying to work out what the right point is and what's in it for the end customer. How do you persuade somebody who's never had a water leak that it's a good idea. Until you've had one, in which case you think this is the worst thing that could ever have happened.
26:51 MG: Yeah, it's just so true of so much in insurance. With my own background and in catastrophe risk, people, if hurricanes don't happen in the last 12 months, they just don't happen. And I think that's very similar to water and fire. Interesting what's happening in the US just now is the company's doing is are targeting the new build. And so what's happening is actually, for the new construction, they're building in valves, I'm presuming the power socket as well and making it near the Wi-Fi. So that if somebody chooses to put a water shut-off valve in there, they just basically fit it directly into the pipe. And actually not only that, apparently the builders are now using this as part of their own testing when they've created the building and they want to make sure there's no leakage, they put their own valve in there. And I think that's, particularly with all the funny plumbing in the UK that we have. Yeah, that's probably going to be one of the access points to get in there for the new build.
27:41 GE: It could be for new build, but if you're insuring Amberley Castle, it's pretty hard to do that. [chuckle]
27:47 MG: The other one I just want to touch on was interconnectivity generally. And as you've built now this full stack digital insurance company, you had the, I guess, the ability to start from scratch. As you looked out there and looked at the components you could use to build a company, how encouraged are you, or are you not, by the ability for these different components to link together either just through a basic API, but maybe a more straightforward level where you can literally get a plug and play of taking the data or the analytics and plug them into?
28:18 GE: Oh, I have a theory that if any sentence you form around this subject has the words data lake somewhere in it, then it's going to be very difficult. And you want to be in an ecosystem. And we don't want to build anything that's already been built. So if there's a great CRM platform out there, don't build your own. If there's great CMS platform out there, don't build your own just take what's out there that's best in class. And it seems to me that there's a lot of bespoke software, and on older stacks, and that is neither built for connectivity, nor is it particularly... Dare I say it? Nor does anybody particularly want it because what you want is to keep people in your own world and not allow them to stray outside of it. And it's hotel California time. And that's the thing you've got to avoid. And if you're in an ecosystem, it's much better if you're running native. But if you're not running native, you build a API in, it's great. And I think it's going to get increasingly difficult to do this stuff cause in the end what we're going to go to is real time data. And you can't go real time in a batch processed world and in a world where there's mantronics and all the rest of it.
29:39 GE: So you have to have interconnectivity. Then you have to look at your core systems, say were they built for interconnectivity? Some of them are great at counting the money, but you ask them to throw data in and out and they're just not built that way. They're not architected that way. And I think it's very hard to do it. And then you look at it and you say, how much of what I've got is built on code that I've got to maintain versus how much is it I've configured? So a good example would be, why do you need to build your own general ledger? There are perfectly good global examples of general ledgers out there that you should be able to connect to. And if you can find one that's in the ecosystem where the rest of your stuff is, so it's built to connect anyway, all the better. It just gives you a seamless, real time experience. And if you don't get that, the frictional costs are horrendous around that.
30:29 MG: So and how do you find the broker platforms and being able to connect into your own technology?
30:34 GE: Well, we haven't yet gone there because we're still building. We would love to connect, but because we believe the product i.e. This digital customer focused, what are your needs? We're going to try and cope with them and give it all to you in a really easy way. If you put that through a broker platform to the end customer, you crush the user experience to whatever the user experience is in the broker platform. And that's not what we want to build. I would love to connect downstream to some of the big broker platforms where the broker keys into what we're doing, gets what they need and they don't have to key it into their own system. That would be really cool. And that's more or less difficult, literally sometimes it's almost impossible and sometimes it's easier, but that would be really cool.
31:24 MG: So the way it works today is the broker would log into the Azur system and put in the limited data they need to get the quote and then you would do it from that?
31:32 GE: Yeah.
31:32 MG: Okay. And any technology companies you see out there that you kind of rate highly for their ability to both be interconnecting, and also actually valuable data and analytics when you've made the connection?
31:44 GE: There's Concirrus doing some really smart stuff with data. There are some really interesting and some good claims companies out there I think doing some interesting stuff there. Some companies talk about the API, but I'm not quite sure how easy it is to actually connect up. So yeah, there's lots of good companies. There's also many, many more that are, I think going to struggle in the new world because they've got legacy platforms and legacy business models that don't suit this new SAS type model. I think if you get the right understanding at the top, any company can do this stuff, but you have to have a CEO that understands what cloud computing is. And if the CEO doesn't understand cloud computing and can't articulate it, how are you ever going to make a decision that's the right decision around this because they've got to engage with it.
32:39 MG: So this has been really helpful. I guess just before we wrap it up, you've recently, very delighted to have you on board as another corporate member of InsTech London. Just kind of interested in what it was about what we're doing that for you and Charlie and others, you made the decision to support us?
32:55 GE: If you sit back and watch what goes on and you just listen to the buzz. We've got a lot of young people in our company. We're 70 people now. And it's a very mixed, very diverse group of people. Lots of energy in there, lots of curious people. And then three or four people come and say, "I've been to one of their events. It was fantastic." And that makes you sit up and take notice. So then you think, "Okay, we maybe ought to be engaging with this and talking to it." And I think you guys are very early on the scene on this and you've got a very good following, lots of energy. And I think...
33:32 GE: I listened to Robin's speech at the Insider Tech conference recently, and I thought he nailed it, and I thought it was brilliant. So why wouldn't you want to be associated with people you think are on the same page and are articulating some of the things that we can see? And then it's really interesting to be around that and to be listening to it and hearing what other people have got to say, 'cause we have not got all the answers, or probably hardly any of the answers, but there's lots of people out there who know lots more than we do, and we can listen to them and learn from them.
34:00 MG: Finally, before we wrap up, what would be your advice to anybody out there who is either looking at entering the insurance world or looking at doing their own startup, from your own lessons?
34:10 GE: Okay, so first thing is, in insurance, I would advise you not to use the language of, "We're going to disrupt everybody." I think there are really well capitalized companies, and I think you should be partnering with them and working with them. The second thing is, find a pain point. Find a pain point that's worth solving and that you know about, that you understand, because that's important. The third thing is, to do something in insurance, in the insurtech world, you really have to understand the regulated side of the business and the technology side of the business. If you only understand one side, you're going to miss it. So you have to marry the two, and I can tell you, that is culturally oil and water. It is really difficult to do. And so it's not easy to do it. If you can make it work, I think there's a huge number of prizes out there for the people that can make this stuff work, and there's some phenomenal companies doing some really interesting things. Risk is doing some great stuff. There's some really smart people out there, and you can emulate what they're doing, but I think the ones that are going to be successful combine both the regulated bit and the technology bit. And I think if you don't get those two elements, it's going to be very hard.
35:25 MG: Great. Well, you're certainly a great... Poster child might be not the right word, but sort of role model at that.
35:31 GE: Poster granddad.
35:32 MG: In how to get out in the industry and actually make an impact and really balance that, sort of make a difference but actually sort of, as you say, not trying to claim to disrupt the industry, but work with the existing organisations that are out there and complement what they're doing rather than necessarily being too disruptive. But at the same time, don't get too caught up in the old ways and get too complacent. So Graham, it's been tremendous. I'll let you get back to all the many things you've got going on, but many thanks.
35:58 GE: Thank you very much. Indeed, it's been fun.
36:02 MG: After that interview, Graham also gave me a chance to sit in on some of the testing they were doing with their new product they are building. So I actually had an opportunity to get hands-on and look and see how the tools they're building are going to work in practice. Really fascinating. You can learn more about Azur Underwriting from their website. We'll also have their details on the notes for this episode. And as a corporate member of InsTech London, we'll be getting Azur up on the stage in the very near future at one of our events. If you'd like to learn more about what it involves becoming a corporate member and how you can get involved in InsTech London or just find out our events, both past and future, you can find us at www.instech.london. And we recommend you do sign up for our newsletter if you're not already doing so to make sure you hear about our events and get a chance to register for those early, because they are booking out now very quickly.