Why AI in underwriting needs to run ‘on rails’: Send’s next chapter with Duck Creek

Matthew: Andy, great to have you back on again. You are CEO of Send. You were founded back in 2017, and most recently acquired by US systems provider Duck Creek, and we had you on the podcast in June 2023. So clearly, for lots of reasons, there’s lots to catch up on. Welcome back. Great to have you. 

 Andy: Great to see you, Matthew. Thank you for inviting me back. And the timing seems fortuitous with what’s happened as well, because we’ve got more to talk about than we would have otherwise. 

Matthew: And yeah, lots of fortuitous things as well. I mean, you’re only a few weeks in since the acquisition was announced. How are you feeling? Has the reality kicked in yet? [01:00] 

Andy: It’s sinking in waves, I’d say. So I’m now actually general manager for underwriting, working for the CEO of Duck Creek, not CEO of Send anymore. That’s a hard one to give up in introductions. I keep slipping to my previous job title. But I’m excited. It’s been a really challenging year. The acquisition obviously generates a huge amount of effort for both teams, the company acquiring and Send that’s been acquired. So to be through that, and relieved to be through it, and excited about what happens next really. 

Matthew: I wouldn’t be too worried about losing the CEO title. I mean, everyone can call themselves CEO these days, but not everyone can sell a company, which I’m sure was very successful. We’ll talk a little bit more about that. But first of all, Duck Creek themselves, unusual name. So quite a few people will be familiar with the name, but less familiar with the company. And I believe they’ve got over 300 clients in the US. Can you just talk a little bit more about who they are and what they do? [02:00] 

Andy: So Duck Creek is an insurance technology provider based in the US. They have over 370 customers globally. And what that means is working with some of the biggest players in the industry, and that includes international implementations across their core offerings — billing, claims and a number of other products that they’ve built to complete the suite. So 25 years plus doing P&C at scale. That’s their business. 

Matthew: And for you and the team — and we’ve known you for quite a long time, so thanks for all your support — obviously these things are always confidential until they happen. But were you actually out there actively looking for a buyer, or was this somewhat opportunistic? [03:00] 

Andy: We were actually out there raising capital to grow. So we were raising a growth round — that means getting an investor to put money in so that we can expand products, spend further, or go to market. What happened, or transpired, as we were having some of those conversations earlier in the year, is Duck Creek popped up on LinkedIn saying, “Andy, do you want to chat?” And that was the beginning of the discussions. A bit of a whirlwind — I think it was around 120 days from that approach to closing the deal a few weeks back. 

Matthew: It essentially strikes me as quite a clever strategy — and as you said, you were out there looking for funding, but as a way of flushing out the market for anyone interested in an acquisition. Because if you disclose that you want to be acquired, you’re into a whole different game. If you just say you’re looking for capital, people will find you anyway. So I’m glad that worked out so well. We’ll come back to talk a bit more about what it means at Duck Creek in a minute. 

Matthew: But just to start off with Send — can you describe the Send client profile as it is today, pre-Duck Creek? 

Andy: So 21 customers. We obviously grew out of the London market, you know that well, but we expanded into North America as well. So our customer base is reinsurers, insurers, MGAs and MGUs. We’ve got customers in London, but in the US, Mexico and Canada as well. So a growing base of customers that do quite different business, but do it in their own unique ways. [05:00] 

Matthew: It’s a really good success story. And you were, I guess, one of the early companies — I don’t think you even coined the phrase, but you were the first person I heard talking about it — an underwriting workbench. And you’ve now expanded into this term orchestration, or orchestration engine. Can you explain what it means to move from a workbench to an orchestration engine? [06:00] 

Andy: The term underwriting workbench has so many different meanings for people. But for us, it was about bringing everything into one place. So underwriters in our customer base are sat at their desk using four or five different applications. They’re handing stuff off to different teams. Bringing all of that work — the data capture, the processing, the integrations — all into one place was really important. But then what we’ve seen, and there’s a convergence in the industry — there are a number of players adjacent to Send, or overlapping with what Send does. Everyone seems to be converging on “engine.” Why that’s important is really about the enablement of AI — that’s the key thing for me. Most successful AI projects in enterprise are based on AI running on rails, not just some app that someone’s vibe coded and managed to stand up in a production environment. You need the infrastructure, you need the governance around it, you need a load of things, and that’s what the orchestration engine brings. So everything we had before, but really a focus on the end-to-end journey, and also the variety of different business flows that our customers have. That’s just diversifying evermore each year that passes. 

Matthew: It’s the kind of strategy that makes complete sense when you look backwards. But of course, when you started off talking about underwriting workbench, the idea of gen AI and large language models — most people weren’t aware of what that means. As you said, the timing has actually worked out really well with that shift to orchestration. [07:00] 

Matthew: People were trying to do orchestration, but it was really hard to do before you started to move into the world of gen AI, I’m assuming. 

Andy: Absolutely. The industry is looking at things in such a different way now than maybe a few years ago, in terms of running tasks autonomously in place of what would’ve been teams of people previously, and that all needing to be choreographed on one platform. That’s the difference we’re seeing. Many insurers in the industry are looking to deploy agents into their workflows, and that’s hard to do. There’s a lot of POCs, a lot of vibe coding going on, but not much that sticks in production.

Matthew: I guess that’s the second half of the brilliance of your strategy — you’ve said this, but it’s worth repeating. For a company to try to start from scratch and build an AI tool, using Claude Code to vibe code it, [08:00] you’d have an application, but it wouldn’t have anywhere to sit, it wouldn’t have any controls around it. You’ve basically hit that at the right time with your underwriting workstation, moving into orchestration and then the AI tool. But I do want to ask about the tagline. I read the tagline from the sale announcement, and I thought it was quite punchy — you’ve hinted at it a little bit, but maybe unpack it a bit more. The combined venture was described as creating the industry’s only agentic underwriting-to-core platform. Can you explain a little more about what’s behind those words? 

Andy: Yeah, it’s certainly punchy — intentionally punchy, I’d say. But what this is about is you’ve got a number of players in the industry. You’ve got a bunch of policy administration core technology providers with massive customer bases, and up until now I don’t think any of them have had a really strong underwriting or orchestration engine story. [09:00] And you’ve got players like Send, and companies adjacent to Send — maybe they’ve come from pricing or AI for underwriting — and some of those are now starting to move towards orchestration engines, some of them are now playing into the core system space. But core systems are really hard to get right, and underwriting is also quite hard to get right. So what you’ve got with this agentic underwriting-to-core platform is what Duck Creek have been building over the last couple of decades — their deep knowledge of the industry, what core systems need to do — together with what we’ve done at Send, which is we spent the best part of a decade building out for underwriting. So bringing those two things together, we think, is an industry first — to have best of breed in the same place. 

Matthew: And for your clients — your existing clients — now you’ve got access to Duck Creek, notwithstanding it’s only a few weeks in, how soon will they be able to start getting the benefit of the Duck Creek technology, combined with what they’re already doing with you and Send? 

Andy: Well, a lot of the customers are already asking what’s coming next — that’s the big question. I think we will be taking more AI to our customers. Duck Creek’s had a really fantastic team over the last year or so building AI expertise, on top of an established, scaled platform for core systems. We’re going to leverage that and bring it to our customers. But also, we’re going to bring a really strong integration story, in terms of Send as an underwriting core with [11:00] policy administration and billing on the Duck Creek side. That doesn’t mean we won’t sell our platform standalone — that’s always going to be a thing. Most of our customers, in terms of Send, are not Duck Creek customers. But we’re going to have a really strong integration story, so that the definition of what an insurance product is that sits in Duck Creek Policy?]</mark> today will be able to synchronise into Send, which is going to lower implementation cost and drive a lot of value for our customers. That’s the plan. 

Matthew: And on the flip side — from Duck Creek themselves, they clearly benefit in that they can sell their technology into your clients. But what about their existing clients? As we touched on earlier, they’re not that active in the London market — can you talk a little bit about what Duck Creek looks to get out of this, beyond your current market space? [12:00] 

Andy: Yeah, I think Duck Creek were interested in Send not just because of the orchestration engine for underwriting, but also because of our London market knowledge. It’s a core platform that’s not hugely deployed in the UK, certainly not the London Market, and they wanted a base of expertise to help expand into that space. For the existing Duck Creek customers, we’re already starting to talk to them — getting in front of people, showing them Send technology, helping them understand how that can work with their existing systems, with the Duck Creek products they already use. I think there’s a lot of excitement around it, a lot of interest, and my diary keeps filling up with demos and discussions with existing Duck Creek customers. [13:00] There’s almost been this pent-up demand for getting something in the underwriting space. All of the insurers out there will be looking to their core systems providers saying, “When are you going to fix this problem? When are you going to have something in that space?” So, pushing on an open door with some of these discussions as well, which is exciting times.

Matthew: Well, it’s fantastic. And we were delighted to work with you on a report recently, talking about orchestration — written by real people, not written by AI. Is there anything in that report that struck you as particularly impactful? We’ll put a link in the episode notes — people can download it from your website or ours. But is there anything that struck you in that report worth talking about? [14:00] 

Andy: I thought two things — one I’ve kind of covered already, which was, you need to run enterprise AI on rails. That transition to an orchestration engine — why is that important now? It’s the platform on which you achieve success with AI as a business. The other thing I found really interesting, and such a good topic of discussion, is the workforce in transition. You’ve got this change going on where underwriting doesn’t get simpler for you — there’s more complexity, more data, more distribution channels, it just seems to get harder. And meanwhile, you’ve got this brain drain occurring where — I can’t remember the exact percentage — a substantial proportion of the underwriting workforce is set to retire in the next [15:00] 15 years. So you’ve got this dynamic where AI is changing everything, and a lot of the knowledge on which everything is based is in a lot of people’s heads, about to go out the door. There’s this kind of existential challenge for these insurers on so many fronts. For me, getting good at AI quickly, leveraging technology so that you’ve codified how you get stuff done in your business onto a platform, rather than relying on a mental checklist and knowing that you’ve got to send something to this other person if something particular happens — getting that codified onto a technology platform is a really powerful way of making sure you don’t lose that knowledge around how things get done. And that’s what we’re seeing with customers — they universally want to deploy their whole portfolio onto Send, because that means they get everything done in the same way. They’ll say, “Well, we work differently in this underwriting team to this other underwriting team,” but they’ve all got common things they have to do — they’ve all got to file a [16:00] report for finance, or do a sanctions check, or so on. There’s a lot of activities that I’d describe as cross-cutting. So for me, it was that piece around the workforce transition — that was the thing that stood out. 

Matthew: You’ve just given me a great idea — a vocal history of the underwriter. You know, they do these vocal histories of disappearing trades, or people who saw military service — you could do one for underwriters, record what they’re saying and then extract it using AI. Though I’m not quite sure how many would be willing to share that. But I do have a question for you on that changing workforce. One of the things I’ve heard talking to people is that insurers have shifted from a scepticism about the power of AI and what it can do, to now looking at their own teams and asking, “Should we build versus buy?” Because now with AI out there and Claude [17:00] Code and all these other tools, you can do more. What are you seeing in that space? And maybe some advice for anybody at an insurance company listening who wants to understand how far they should go with their own team, versus at what point they need to bring in an orchestration engine because they can only go so far on their own? 

Andy: We would say this as a software vendor, but I do believe it as well — buy the commodity and build the differentiators. Where we see really interesting things going on in our customer base is where they’re doing that internal build, but doing something really unique. They’re doing something that’s a true differentiator — using data in a certain way, automating part of the process in a unique way. That’s where the value is for me. I don’t think all the insurers are about to start building their own core systems, because there’s an overwhelming demand for building [18:00] more stuff — there’s never been more software engineering roles on the market for hiring. So there’s this demand, but AI’s just reduced the cost of doing software development. I’d say do something that’s really useful and unique with it, rather than — no one’s going to go build their own version of Office 365, are they? You’ve got to draw a line somewhere. But as software vendors, we’ve got to keep up with that, keep ensuring we’re bringing the value, but also enabling our customers to build those differentiators and deploy them on the platform. 

Matthew: And that uniqueness plays back to London’s role as the insurer of last resort, or for things that are so complex, or excess and surplus, where many of the best underwriters and their actuaries would have been using spreadsheets to do their analysis and know clients well. Now they’ve got the tools, but it’s just different tools  

Matthew:  And then, switching a little bit now to the founder role — big question for you. If there’s one bit of advice you’d give to yourself looking back on this process, that you could also give to other people — something that’s either helped you be successful in this sale, or something you might have done differently. What would be your advice to somebody else selling a business at some point? Something you really have to get right. 

Andy: I’m going to say something that’s more around building the business, so that you build something valuable. But I think the thing that’s been remarkable for me is having a bunch of people around me who are aligned — not completely, but somewhat aligned in how I see the world, what matters to me. That was important right from the outset, for me and my co-founders, Matt and Ben, that we’d have people who wanted to get things done, wanted to do the right thing, but were ambitious and wanted to build a business as well. You can find people with really strong functional skills — you can test people for programming languages, or whether they can do particular finance tasks. But you can’t easily test for people who actually care about what you’re doing. That’s hard to find. But probably a really key thing for us was assembling a team full of people who actually think things matter for our customers. [20:00] 

Matthew: And if I can add something as well, just on that — one of the areas I feel you’ve been very successful in is seeing marketing as a strategic role within the company, and Sarah joining you, someone who really understands the technology. Frankly, because we talk to lots of companies, someone who actually knows what’s going to support the business and the [21:00] company is just incredibly valuable. I’d definitely call that out to anybody looking for advice on who to bring into the team. We’re going to come back to business in a moment, but just on the advice front — I’m always intrigued, particularly as we’re heading into the holiday season, whether people have a book or a podcast they can recommend, something that’s been useful for you? 

Andy: There are two things I’d say, if I’m honest, that are my go-tos for podcasts. One is nerdy work stuff — Invest Like the Best by Patrick O’Shaughnessy, which is how investors see the world and investment opportunities. But it goes deep, and there’s lots of interesting storytelling, success stories and failures — I like that. But perhaps a bit more basic is Desert Island Discs. I always go back to it, and there’s such a rich archive of them, going back decades. And [22:00] what I like about it is that you’ve got people being — I don’t know — put into a sort of situation where they’re channelling a memory, and they’re talking about something, and it catches people off guard. It’s lovely. So those are my two. 

Matthew: I haven’t heard of Invest Like the Best. And on Desert Island Discs, I noticed Christopher Nolan was on there recently — so for anybody who’s seen The Odyssey and wants to understand what’s going on in his brain, or what music he likes, that’s more Desert Island Discs. Definitely recommend that. Now, Andy, those are two good ones for the holiday season. And finally, for people who want to know more about Send and Duck Creek, what’s the quickest way to get connected with you or your colleagues? 

Andy: Well, you can always reach out to me on LinkedIn, and check out our website and our LinkedIn pages as well. We’re an approachable bunch, [23:00] so please just say hello if you want to find out more. 

Matthew: They are — I asked somebody once at one of the dinners we did with you, “How do you know Send?” And they said, “I listened to the podcast and now I’m a customer.” 

Andy: I know who you’re talking about as well. 

Matthew: Hopefully still a customer. Andy, you’ve got a lot going on absorbing the new owners. One final question for you — having worked with Americans for a long time, I have to ask: where are Duck Creek based? What part of the US? 

Andy: They’re remote-first. Their head office is in Boston, but they’re quite spread out across different states. 

Matthew: Good move, first of all, getting an East Coast US company to work with, just because of the time difference. One thing I learned working with Americans is you do need to remind them that work stops in the UK at about 6 o’clock. So please arrange — certainly for the team, I’m sure you’re very diligent, but please don’t arrange calls after [24:00] 1 o’clock East Coast time. 

Andy: It’s as though you’ve seen my diary. But no, things are settling down now, so it’s good. 

Matthew: Right, get it right early, because then they learn. Anyway, Andy, it’s been a pleasure to talk to you. Congratulations — I’m really pleased to see the news. Thanks for sharing that with us early as well, and hope to see you in person soon. 

Andy: Thanks, Matthew.