Podcast 52. Amrit Santhirasenan: CEO & Founder of Hyperexponential - modelling reimagined

Twitter icon
Facebook icon
LinkedIn icon
Podcast

For this week's episode we're back in the Lloyd's Lab talking to Amrit Santhirasenan. Amrit is a qualified actuary and having become frustrated by the lack of available tools he could use to run his actuarial models, he founded Hyperexponential. Previously head of pricing and analytics at Tokio Marine Kiln, Amrit founded Hyperexponetial (or HX) in 2017 and he already has clients and revenue. Matthew finds out from Amrit how his time in the Lloyd's Lab Cohort 3 has been accelerating the company's success in finding supporters and clients in Lloyd's and beyond.

Look out for our write up of this interview coming soon, and save time from taking your own notes by reading the transcript below.

The Instech London podcasts continue to be supported by Insurance Insider, one of the best and fastest providers and insights of information for the global commercial, speciality and reinsurance markets. Get your free copy here:  http://campaigns.insuranceinsider.com/instechlondon/

Listen here to podcast 52. It is also available on iTunes, Spotify and Podbean.

......................................................................................................

Transcript for this podcast

00:00 Amrit Santhirasenan: There's a word which I'm a big fan of, and it's called scientism. And it's a belief that if you've got a good formula in a number, you've got the answer every time. Working in this industry, I've learnt the hard way, having been a guy who loves building a complicated formula as much as the next guy, that that doesn't mean anything as to its predictive power.

00:19 Matthew Grant: Welcome to the Instech London Podcast, Matthew Grant here, and for this week's episode I am back in the Lloyd's Lab, this time talking to Amrit Santhirasenan, CEO and co-founder of the Hyperexponential or HX in short. Amrit is an actuary and founded HX after years of frustration of not being able to find a decent platform for his own actuarial models. He launched HX with his former employer as his first client and today, the company is generating revenue. He is another member of the Lloyd's Lab and Cohort 3.

00:58 MG: Amrit, thanks for inviting me back in to the Lloyd's Lab. So it's been a week now since Lloyd's blueprint one came out. I think you said you've managed to move the consultant side, so it's a bit less hectic here, but I'm sure it's buzzing with all of the recommendations in there, many of which I think are going to play into what you're doing.

01:16 AS: Lots of exciting things coming out of it which definitely play into and influence the way we want to go as a business.

01:21 MG: Good. Well, let's jump into what you've got on your website, which says that you build tools to help people price from weird and wonderful data sets, let's start off with that, and find out a bit more about what you're doing.

01:31 AS: Absolutely. We refer to weird and wonderful data sets 'cause actually those are what make up the majority of the data that drive what our clients use to make decisions. So if you're a specialty insurer and you're trying to price Tesla or the PA cover for a football team, the sorts of data sets you get, they're going to be a little bit strange, they're probably going to be small, they're going to be awkward, and you're going to have to do quite a lot of work to put them together to use them to drive a decision. So they're weird and wonderful in a positive and a negative way.

02:02 MG: Which I guess is at the heart of what Lloyd's does itself is a market for underwriting complex risks, you're essentially just helping them do which is one of the main areas that Lloyd's is offering up to the market. And one of the major themes coming out from Blueprint One is the area of the complex risk exchange.

02:20 AS: Yes, yes, absolutely. The backbone of Lloyd's is specialty complex risk. It's what it's famous for, it's what occasionally, people when they do hear about us in the press, it comes for those sorts of reasons, 'cause it is the kind of bastion of complex difficult to place risks.

02:38 MG: And you're yourself an actuary. You've founded a business, you've got a couple of other actuaries here, it's quite a potent force. You started up in 2017 coming out of Tokio Marine Kiln, how has it been going in the last couple of years?

02:51 AS: Great. Yes, absolutely. We are a bunch of actuaries, we build actuarial pricing software. So to some extent, it was inevitable that there would be a few actuaries in the business. We're not full to the brim of actuaries, we're a technology company, so predominantly engineers, but Michael and I, we are both co-founders and we are both actuaries. Now we're a real business with clients testing our product, giving us challenges, giving us new things to do and the market as you alluded to earlier with Blueprint one and the future of Lloyd's strategies moving fast as well. So it's great, it's exciting, it's certainly keeping me very busy.

03:24 MG: Well, congratulations and having real clients is certainly one way distinguishing yourself from everybody else who's out there. So we're in Cohort 3 of the Lloyd's Lab at, this recent cohort's been going about six weeks. So what was it that brought you with an existing organization, you've got presumably quite well connections into the London market, just given where you came from, why did you choose to join the Lloyd's Lab?

03:46 AS: We're also really interested in thinking about what best practice pricing looks like, and if you're going to do that sort of thing, it's really worthwhile to engage with the Lloyd's. So one of the things we wanted to do was use the lab to put a structured framework around collaboration with the corporation, that was really, really important to us. Having the lab here, it's been absolutely fantastic for kind of capitalizing those connections with Lloyd itself. And also, it's provided us with a hub to work with our prospective clients and Lloyd's together. So it actually was a very obvious place for us to come because actually what it does is it short circuits lots of the kind of inevitable and administration in dealing with the larger organization and it's certainly done that.

04:26 MG: Good. And so you get the introductions, you get some mentors, you say it's going well, are those starting to turn into commercial relationships that will continue once you get outside of the lab?

04:36 AS: I really hope so. I think we certainly held our end of the bargain up. We set ourselves three very clear objectives, one was to demonstrate the power and agility of our system, so to get the system, our products from an idea or from an existing older system, Excel or another one of our alternative systems that our market uses, into our system and demonstrate the speed and agility. We've already done that with at least two of our mentors. So, demonstrating that, and therefore we're already at the stage halfway through the lab where actually we can now take it beyond just an idea and move it into a commercial relationship. We had a couple of other objectives working on best practice pricing and collaborating with our fellow cohort members, both those things are going really, really well, as well. But to cycle back around to your point about moving forward commercial relationships, I'm feeling really good about it. They've been really, really, the people who have engaged with us from our mentors have been incredibly positive, very supportive, and I'm feeling good about moving things forward.

05:28 MG: Now the way you started the business follows some of the conventional wisdom for how people should start businesses, which is you're working at a company, you're trying to get something done, you can't do it. You say, "I can do this better myself." And you went off and you started up a business. So what was the sort of problem that you had, that you couldn't solve, that you're now solving with HX?

05:48 AS: Oh, really good question. I would say, the single biggest challenge we had was speed and agility of building models, getting them into the hands of our stakeholders, the people that we serve as actuaries. So when I say "We" now I'm going to zoom us back a few years and imagine that I was the head of pricing at Token Marine Kiln which is where I was before here. One of the biggest challenges we had, and I've learned this through the 13 years I was in the market, is that actually, specialty pricing, it changes, it changes fast. We work in a really dynamic interesting, fast paced market, and we could not find the tools that allowed us to keep pace, to get the analytical tools into our underwriters hands at the pace that the market was changing. So, our raison d'être is to build a product that's designed to match the complexion and the nature of the market itself. And so if I had to put one word down, it would be speed. And that's something that we're really, really proud of, and as I said, circling us back round to the Lloyd's Lab, motivated by the rugby world cup, one of our model developers built a model in 32 hours, to get a model into the hands of one of our mentors before he left. So, to me that's the ultimate in vindication, is that if we can do that sort of thing.

06:57 MG: Good. And did he underwrite on the back of the model?

07:00 AS: Good question, we'll have to find out about that.

07:01 MG: Find out when he comes back.

07:03 AS: Indeed.

07:03 MG: So we talk about building models, I'd just like to just take a step through how AJAX actually operates. So you're essentially building the tools to help the actuaries use, I'm assuming it's like lost data from their own experience. Are there other sources of data they would use to build those models and then sit on your tool?

07:23 AS: Absolutely. So, the way we see this is that we want to be a platform, an end-to-end platform, for the actuaries in our client's companies to build models, to deploy them through to underwriters, so our underwriters can use them to capture data and to generate insights, and then to handle all of the other bits in between. So in terms of answering your question about where the models come from, they're our client's models, they're either existing models they've got in spreadsheets or alternative platforms, they're parametrized from lost data, from underwriter judgment, and increasingly now, and this is one of the things that relates to one of those objectives I mentioned with the Lloyd's lab, is they're pulling data from third party sources. So if I plug the friends from Insure Data or Climate Cell, they're using the data from there to give them extra insights. The way we see our tool is to help pull data from all sorts of sources, historical data, modern API-centric feeds, underwriter judgment, put them into a big pot and get them out there to use and to test very, very quickly. So it's the full gamut of data sources.

08:24 MG: Right. And I suspect it's a two-stage process, isn't it? So presumably the actuaries and your clients who are your typical users, they would find the data either from their own losses or from third parties or a combination of both. From that, they'd put figure out how they want to build their pricing tools. They then put that into AJAX, which then allows those to run that in an underwriting scenario. It's not pulling in data dynamically from the other sources. So you're not sort of pricing on the fly you're actually building the tools in the background.

09:00 AS: Really good question. So just to help you and do a bit of a plug, our product it's called Renew, and that's the platform where this happens. Renew has the capability to pull dynamic information in on the fly as well. Some people have static models, which exist just pulling data in from their existing data repositories to generate prices. We have done proof of concepts and we do have some models that we're working on right now where data sources might be pulled from anything from like a Google newsfeed or some of the third party data sources. Again, like I say, some of the APIs that we mentioned earlier. And therefore to add that element of dynamism in the market, and again, that relates to what we were saying about changing fast.

09:37 MG: And what are you seeing as you look out over some of these complicated risks in the specialty business, in terms of sufficient data being available to do the underwriting, 'cause by their nature, these risks come into London 'cause they're difficult to price. Although everybody recognizes the value of data, it's still incredibly difficult to be able to find the data that people actually are comfortable using to price. So is that kind of going to be the next barrier for you now that you've built the tools, people are using it, but actually now can they get the data they need to be able to do the underwriting?

10:07 AS: Yes. So that's a great question, an absolutely great question. And actually touches on exactly why we're here. There's a huge challenge in getting data. I say, I was at Giro, one of the most scintillating actuarial conferences in the world, I was talking about a few weeks ago, and one of the things I said is that, if you're waiting to get the data from the broker or in the submission, you're already behind the curve. So people are looking out there. They recognize this challenge of getting data from any... Getting the right sort of data to do the analysis. One of the things we wanted to do is make it easier, to make it easier and we use kind of modern open-source technology to get data from a variety of different sources. I don't think that's a barrier that we have to overcome next. I think that's a barrier that we're at right now, but our... That our market faces right now. But our product exists to make it easier to get data from all sorts of sources. It's a huge part of what we do.

10:57 MG: Good. And how do you fit into the underwriting workflows? So you're essentially creating another tool that people have got to figure out where in the workflow does that fit, who runs it, does it become another barrier or can you actually make the whole process more efficient? Which is where Lloyd's, for one, is trying to go. How do you stop yourselves becoming yet another choke point or a narrow set of skilled users versus you actually are helping reduce the overall cost of the transaction?

11:26 AS: That's a big, big challenge. That's not a challenge that's a function of the system, it's a challenge of the kind of work that goes into establishing the pricing workflow. It's something I've talked about at length in the past, again at some of the presentations I've done, where having an unclear underwriter workflow is a huge source... It's a huge source of a pinch point. That's not something that is necessarily related to our system or any of our competitive systems. We do spend a lot of time at the very beginning engaging with the clients and on really challenging them on whether they understand clearly their underwriter workflow. It's almost a sort of thing that the system doesn't cause the blocker, it's actually just not having the process in place to actually understand where we sit. To answer your question as to where we can sit, well, actually, we're a system that can... In our base functionality right now, as of right now, we can pretty much deal with all the classical actuarial models that people build.

12:22 AS: As a result, we can sit in the workflow where the underwriter is doing the pricing. We can sit in the workflow where an underwriter fills out a little bit of data, we pull that data from a different system, and the actuaries does a slightly more technical analysis. We can fit into any part of that workflow. And one of the things we said when we were pitching for the Lloyd's lab is, we're absolutely an ecosystem play. We are not a monolith, we're not someone that's trying to eat up the entire value chain, sell vendor lock-in for ourselves, that's not something we're trying to do. So we will fit anywhere and we will push and pull data from wherever we need to and trying to be API first, API-centric in our approach, which we have succeeded in and it allows us to do that.

13:00 MG: And where do you sit on this view about legacy because that's often given as the reason why these tools can't be adopted. Either because just technically it's difficult or impossible or it's just a whole different way of doing things. But for you to be successful, yes, it sounds great to have the APIs and certainly we're moving into a new world or platform, but are you able to overcome some of the resistance, be it both legacy thinking and legacy technology, to get yourselves deployed?

13:25 AS: That is a very important philosophical question that I think if I knew the answer to, I'd probably be giving you it from my yacht. So what I would say there is the way we go about this as an early stage... A relatively earlier stage startup in a market where there's still a huge amount of it being untapped in old systems, there are people who are very API-centric and who are trying to do that right now. We have a very natural and logical appeal to those. The slightly more traditional insurers, I'm not going to sit here and pretend that we have all the answers to every single solution. They do find API-centric platforms harder to deal with. However, to go back to the philosophical point, because of that a whole ecosystem of robotic process automation and other tools are popping up to make integration with legacy systems much, much more realistic and much more doable. Mercifully that means we don't have to solve that one ourselves. And as a result, can we integrate with legacy systems? Yes. Is it a high priority for us at this point in time? No, but we're working on... And in fact, in the Lloyd's lab there are people who are looking at RPA as a solution to that.

14:32 MG: And on the high priority one then, I guess it also takes me to another question on this, which is if organizations can find ways to better price new products, new risks coming into the market. Most people now have heard about the switch from 40 years ago as 80% of risks were tangible physical assets, now 80% are non-tangible. It's still really hard to price many of those, Predicat are here in the lab doing that for liabilities. Cyber is obviously one of the bigger ones. Is that an area that you're thinking about specifically as where you can really add value because if you've got the pricing tools and you're offering a new product or you're enabling other people who have got a new product, who cares about legacy? You're bringing new revenue into the market.

15:14 AS: Massively so. I've just come out of another meeting in a demo with another very large Lloyd's syndicate, where we made the joke. It's a cyber model that they want to look... Are working with us on. And I always make the joke that, do you underwrite risks that are coming to market in 2022 like it's 1996? We shouldn't be doing this. It's an area of intense interest to us, we're absolute believers. Again, I would say this as a former head of pricing, that good quality pricing and analytics has a direct impact to the bottom line. We've demonstrated that in our past lives as actuaries, as leaders of actuary analytical teams. To answer your question, is this an area that we're focused on that's a high priority? Absolutely. And we do believe that people who invest in this now, just from a fundamental point of view, the Google guys who said, "More data is better." You can't apply that in a completely reductive form, but do we believe that in areas like cyber, IP intangibles actually... Can you get a leg up by doing that first? Absolutely.

16:11 MG: But do you also have to be slightly schizophrenic with your actuarial background? Because actuaries like to see proof and data sets and risk that has to be measured. Even looking at some of these tools to move into the non-tangible areas, many of which they haven't had losses or have losses so rarely that it's very hard to model. From a personal point of view, I guess you did a start-up so you've already by definition you're probably a fraction of a cent of any actuary out there, so probably it's not a problem. But is that something you find you got to reconcile with yourself?

16:40 AS: No, it's not. I probably... As you rightly said, I probably fall into a slightly different camp of actuaries. There's a word which I'm a big fan of, and it's called scientism, and it's a belief that if you've got a good formula on a number, you've got the answer every time. Working in this industry, I've learned the hard way, having been a guy who loves building a complicated formula as much as the next guy, that that doesn't mean anything as to its predictive power. And actually, it fits to our core that actually... I suppose I should centre this on your question, do I feel schizophrenic? No, because I really firmly believe that we don't have all the data right now to answer these questions. As a result, experimentation and rapid iteration and finding out what works and doing more of it and just doing less of the stuff that doesn't work is going to be the secret. So, is this me being... Do I... Am I schizophrenic in it? No, I'm very, very clear-headed that less is more in certain areas.

17:34 MG: You're an actuary for the latter part of the 21th century, that's...

17:37 AS: Maybe, maybe.

17:39 MG: Excellent. So, are there... Anything that's standing out specifically there in that area? You know, you've got a lot of things... I suppose the risk, building what you're building, is it's a wonderful tool but can people really focus specifically on what they can use it for? Of all the people you're talking to or maybe actually you've got clients paying you money, are there anything specifically that stands out as where people are starting to use you that you can talk about?

18:06 AS: Completely. The biggest areas that we're being used for are where people need to deal with... I'm going to sound like a broken record, but that's okay, 'cause I think the record's got a good song... Is where people need to build things quickly. We're working with one or more clients, they've got big, big challenges to get up and running, get models out to systems really, really quickly, and that's where we really can stick head and shoulders above our competitors. Things that typically take tens to hundreds of days to get built, in a system like ours we can do much, much more quickly.

18:35 AS: I am an actuary, so I'm really loath to give performance differences 'cause it will be a small subset of samples, but that's a big area. In terms of product lines, we're agnostic. We're being used or being... On anything from classical liability models, through equine, through energy, through to some of the more esoteric... We've got a project down the line to work on an IP model that's coming through. In terms that side, no. The characteristics are people who want to pull data from lots of sources and they want to make changes and developments really, really quickly.

19:08 MG: Yes, that comment about speed. We're hearing again and again and again. I think the mantra for insurance is not fail fast, it's actually succeed fast.

19:16 AS: Yes.

19:17 MG: You're part of that and so it's interesting. You're saying that the data itself is... Or the models themselves are less critical, it's actually people that want to deploy it and they don't want to get bogged down in friendly old spreadsheets.

19:28 AS: Massively so. One of our mantras is, "Let HX pull the technology risk away and let you focus on the insurance risk." That's why insurance companies setup. I learned this, having been a software engineer at university, that was one of my main reasons I ended up with the engineer-itis they call it. The delusion that it's... I'll do this, I've got this, I can build the system, which is wonderful because now we have done it. Actually turned out it wasn't so much of a delusion, but one of the key things is absolutely focus on succeeding fast and getting something out there.

20:00 MG: And the insurance industry loves acronyms. We can talk about three letter acronyms or of some four letter acronyms out there. You were talking to me earlier, and you've got a five letter acronym.

20:06 AS: I have. Yes, Practically Effective Pricing and Specialty Insurance. So, I seeded this went out there at Giro in Edinburgh a few weeks ago. Absolutely, it's something that I've always been very passionate about is building useful, impactful pricing models. I started off with practical impactful pricing models, but PIPSI didn't have quite the same ring to it. It's something that we're working on. I'm going to be doing a presentation at the Lloyd's Lab, doing the same presentation I did a few weeks ago to get it out to the market more broadly. We're seeing actuaries and data science and analytics teams becoming much more results-focused than they have been done in the past. It used to be much more... My boss, my former boss uses the phrase, "The yellow cell is in the spreadsheet that gave the answer." It used to be about that, whereas now it's actually, how are you making a business better? We're expense constrained, we're under huge time pressure. People are saying, "Do more with your data." And that's what PEPSI is about.

21:04 MG: Brilliant. Well, I'm sure people will remember at least the acronym name, even though they can't actually take it all to bits and repeat it. So what's next after the lab for you?

21:14 AS: A really good question. I don't want to say it's back to normal business. We've actually... The lab's been a fantastic catalyst for us. I think pushing further on helping the market or working with the market on Pepsi to move things forward in that side of things, growing our business, working on sales. One of the wonderful things about the market we work in is actually, with our existing clients, there's huge frontiers of unexplored territory.

21:40 MG: But it's... Ultimately, it's clients product delivery?

21:43 AS: Yes, absolutely. And actually we've got a lot of innovation and experimentation that we want to do over the next year. HX has evolved a lot. We are absolutely very, very client-focused and they've got some fantastic ideas, and that's actually the bit that really gets us out of bed, is helping them do new things.

21:58 MG: And so, you've got a lot going on, but you've also been good enough to come along and see us at an evening at InsTech London for our events.

22:04 AS: I've been coming for years.

22:05 MG: So, well, yes, thank you. And so, what is it that brings or keeps bringing you back to what we're doing?

22:11 AS: Lots of things. I think one thing I've learned having worked in technology for the last couple of years in a dedicated fashion, the power of community is something that's tremendously valuable. I'm a huge believer in it. We went originally to your events a little bit somewhat speculatively while HX was a twinkle in my eye, looking at what other guys were doing. It was hugely valuable for us and inspiring actually to see the other guy's success and the tribulations that they go through but what it means for them. So originally it was a little bit of inspiration. Over the last couple of years, it's been a little bit more about getting on top of the market, understanding a little bit more about what the market is doing. And in the future, I think it's probably even more about engaging. Maybe you might see me up talking at 150 miles an hour on stage at some point. So maybe we're working our way from the back through to the front, I'm not sure.

22:57 MG: We're definitely going to get you on stage but the challenge is to get 30 minutes of conversation into six minutes on stage.

23:01 AS: Yes.

23:02 MG: But yes, you've got lot of great things to talk about.

23:04 AS: Thanks.

23:05 MG: Good. Well, that was really, really helpful. If anybody wants to find out more about you how do they find you?

23:11 AS: Loads and loads of different ways. LinkedIn is the natural place in the modern social media eco-system. You can go to our website, if you can spell it, hyperexponential.com. You can reach out through the Lloyd's Lab. You can find us on Twitter. You can find us in loads and loads of places. So lots and lots of different ways.  I'm very regularly in the market and in the lab. You can find me in a coffee shop in a square mile radius with little difficulty.

23:39 MG: Good. No excuses then for not finding you and actually just talking about the benefit of London, I was having a conversation earlier on with somebody who'd been at Vegas. And they said the thing about London it's like Vegas every day, you've got 7,000 people or more all doing things in technology all within about 10 minutes reach of each other, and it's a great benefit we have. It's that community as you say, it's talking about technology, and increasingly, it's speed and yes, prove you can do it.

24:03 AS: Absolutely. Innovation is about delivery. Otherwise they're just ideas. Really exactly, doing it. It's really, really true.

24:10 MG: Good. I'll let you get back to all the things you have got to do. Good luck at the end of this cohort when you've got your final presentation and we look forward to catching up, catching up soon.

24:18 AS: Thanks for having me. I really enjoyed it.

Listen here to podcast 52. It is also available on iTunes, Spotify and Podbean.