In this episode of Making Risk Flow, host Juan de Castro sits down with Cassandra Vukorep, Chief Data and AI Officer at Lloyd’s of London, to explore the impact of data and AI on commercial insurance. They discuss Lloyd’s role as an ecosystem platform for over 100 syndicates, the importance of data standardisation, and efforts to streamline operations through the Core Data Record (CDR). With only 8% of insurers adopting AI across multiple functions, Cassandra highlights key challenges, including legacy systems and process redesign.
They also examine how AI will become a competitive differentiator in the coming years. This conversation offers valuable insights into Lloyd’s digital transformation and how insurers can navigate the complexities of AI adoption to drive efficiency and innovation in the evolving insurance landscape.
Listen to the full episode here
Juan de Castro:. Welcome to another episode of Making Risk Flow. Today, I've got the pleasure to be joined by Cassandra Vukorep, who's the Chief Data Officer at Lloyd's. Cassandra, thank you so much for joining me today.
Cassandra Vukorep: Thank you for having me, Juan. And I probably didn't tell you this earlier, just recently, I am now the Chief Data and AI Officer at Lloyd's of London. So I'm a little bit excited about that.
Juan de Castro: That's fantastic. Congratulations. But that is the first step to start making progress on AI in Lloyd’s. I'm sure we'll touch on AI in a few minutes' time. But can we start with you giving us an overview of your background? How did you end up with this Chief Data and AI Officer role at Lloyd's?
Cassandra Vukorep: So I don't have what you would call a linear career. It was really interesting. I had a young woman sit me down, early 20s, just graduated a couple of weeks ago. And she asked me, she goes, how did you plan your career? Like, what was your next step? Did you think about this? Did you always… I looked at her? And I was flabbergasted. I'm like, d you think I planned this out? Shocking. I'm like, no. But I do think the nonlinear aspect, I started in wealth management. I did my MBA in the UK. And then I went into IT services, big outsourcing company called Infosys. So really learned that kind of that area of the business. Then into big program change around architecture, digitisation with the big insurance companies. Lots of business development sales experience throughout my CV. And then I landed on the Blueprint 2 data program to run, to get the standards in the MRC across the line. And that's how I ended up at Lloyd's as the Chief Data Officer. But it's all those elements of IT, business development, business knowledge, actually working almost like a broker on the front end that helps you understand where data should be and move it from being a tech function to kind of a business enabling function. So at least that's what I tell myself.
Juan de Castro: At the end, in the type of role you've got, it's not just about the theory of the role of data and AI. It's more about having the strategy and the theory. But then how do you drive change? And I guess your background in Infosys and others really gave you that balance between the strategy and the execution.
Cassandra Vukorep: Exactly. So yeah, all the pieces of the puzzle that lay together have ended me up where I am today. And I feel it's nice because data covers everything. You're not a silo within a corporation. You affect every element of your corporation. So having a little bit of understanding of different elements of what operate really help you enable and be able to make decisions that are business focused and not saying a little bit narrow in your decision making.
Juan de Castro: And I know that you've mentioned in the past, just as a quick side note on that career advice to a graduate. As you said, many people think that successful people have the whole careers planned. And I think one of the things that resonated when we spoke in the past is what you've really done is you've jumped to interesting challenges and opportunities. I think that's probably been the theme throughout your career, right?
Cassandra Vukorep: Yes. And it's always, try and don't be scared to fail. I think the only failure there is is not trying and staying still. So I encourage all of anybody that I mentor is going, if you want it and you want to go after it, don't be scared about what you can't do. Be focused on what you can do. And if you don't get it, when one door closes, another one opens. But that's the best career advice that I can give any young person is just don't be scared to try because failure is just a part of life. We all fail and then we grow from it.
Juan de Castro: Exactly. Well, there you go. You've got a bit of career advice as part of the episode. So all those different roles that you've had and you've worked for, I mean, mostly focused on financial services and insurance, but also exposed to other industries. What do you see the insurance industry and specifically Lloyd's at the moment?
Cassandra Vukorep: Well, insurance itself, because I did come from the commodity side. So the acceleration of change when you're looking at your P&C businesses is a lot more than when you're looking at your specialty business, which is a lot more entrenched in legacy culture. But really, it has big barriers to entry and it doesn't really have that threat yet to change. And as humans, we do all like, why change? If it's not broken, don't fix it. And we are an industry where we're not being pushed to change. So you would want to change as opposed to a need to change. There is also not that money stick that kind of sticks around that if you had a board level going, we'll need to cut costs, etcetera, etcetera. So for good or bad, it probably hasn't moved as quickly forward as you would have seen other industries that had different economic circumstances that forced them to change. But there is a lifeblood from the future at Lloyd's work, from the data standards work, from the Data Council, from all these other elements that are sitting within the market. People want to progress. They want to move forward. They don't want antiquated ways of working, and they're coming around to understanding that the data is actually at the root cause of all of that. So some of their constraints, but also their liberation, is data, which is nice because it helps people in my position start echoing the things that they need to do to actually change. And it doesn't need to be this monolithic change. They can make incremental changes that will affect their everyday working life. And hopefully their profitability with small changes to how they work their data.
Juan de Castro: How much do you see the pull versus push kind of role of Lloyd's with their participants?
Cassandra Vukorep: Well, Lloyd's, it's an ecosystem, right? There's a hundred-plus syndicates that we represent. They all have different commercial objectives themselves. So they are all very commercial. But Lloyd's itself is a society. And our job is to help the ecosystem of the syndicates, managing agents and the brokers to create a platform where they can thrive and they can be profitable and the business that they write can be profitable. So we've held a really important spot in the ecosystem, though ourselves are not a commercial entity where we're pushed by a board to hit certain profit margins. We do have that view of profitability, as you see in the results that will come out of how we help the market stay profitable.
Juan de Castro: I guess within that context, I mean, when you're dealing with a hundred plus syndicates, some syndicates will be more eager to drive change, some others less so, or just don't have the capacity to adapt to that change. So is that a challenge?
Cassandra Vukorep: That is a challenge because when you're developing your strategy, you have to take in the range from the most advanced and forefront of like they've got AI embedded, they're moving forward, they're doing things quicker, they're tech enabled to syndicates that are still probably CSV, Excel based. And they all write business on the platform. We provide a platform for everybody who has a syndicate and we have to be able to adapt. And you're right, that is probably one of the challenges because easily if you're a commercial entity, you would rationalise out, you go, I'm just going this way. This is my client base. This is how I'm going to do it. But no, we are a platform. We're an ecosystem and we have to support that ecosystem and everybody in that ecosystem.
Juan de Castro: So you mentioned data as being one of the enablers to drive that change. Would you say, what are the top three enablers that you think that are going to be at the cornerstone of driving change in the industry?
Cassandra Vukorep: So one is what we do with data. There's a lot of duplication, a lot of manual entry, processing, etcetera. And it's actually looking, it's connecting data and process together and understanding if you looked at your data governance, if we say data ownership, and you drew everything back and you went, okay, who actually owns the data? Who cares about the data? And then you start designing your process based on that. You're going to rationalise out a lot of manipulation, rationalise out a lot of handholding, and you're probably going to get a smoother transaction and much better quality data, which then will go on to the next one, which will get much better insights, analytics to help you strategically move forward in what you want to do. And then it comes into once you have those pillars stabilised, right? Because you'd have good data governance in place. You'd now have really good strategic analysis. Building out what your data ecosystem and platform is. It's not going to be this overwhelming tech platform that's trying to cater to every single requirement that needs to be there, doesn't need to be there. It's going to be something strategically slick that can help with operational efficiency. So they're all intertwined together, but it's people seeing, I think from the beginning, it's people seeing that data and process are one of the same. And if you'd look at the data, you can fix your process.
Juan de Castro: So it's really data, process, technology. So on the data side, like perhaps for those less familiar with Lloyd's, what are the key initiatives that you are working on or driving?
Cassandra Vukorep: So from the data side, where we can actually make an impact is making it easier for the syndicates to interact with us. At the moment, Lloyd's asks for 96 plus different market returns. And they're in CSV format, Excel format. They come through multiple different ingestion engines. But even the data itself, we're asking the data to be manipulated within the managing agent systems before it's sent in to us. And it creates an industry of operational burden to interact with Lloyd's. And this is probably your commercial aspect coming in. We want it to be more operationally efficient to work with Lloyd's. We want to ensure that the data that's being provided is data that the syndicates or the managing agents actually care about. It's something that they use within their own systems because that will help with our analysis and the purposes of what Lloyd's does. So a couple of the projects that have been running and that have the quarterly annual, so we had the QMA and we had the TPD. So that's preserving and central finance looked at the data that they were ingesting and said, actually, we don't need all of this. Like, this is all we really need to run our lean process and then redesigning the processes. And they went out to the market and interviewed them to make sure if we change this data ask, we're only asking for X, Y, and Z now as opposed to A, B, C, E, D, and E. Is that easier for you? Does that make it better? And that was the response. So we're implementing those two projects. And we're now kicking off with a couple other functions, who own big market returns to make it easier for the market to interact with us. And also to bring transparency. We learned from the reserving project, even though we, obviously, we need data to run Lloyd's processes, right? We're a platform. We need your information. We need the data. But there was a clarity and a transparency of what each data point was being used for, why it was needed. And therefore, our clients, who I consider the MGAs and syndicates, understood. They're like, oh, I'm happy to provide that to you because, yeah, we really know that you need that in order to operate and keep the platform going. So that's not a problem. People are resistant when they're going, why do you need this? I don't understand what you're doing with it. So bringing that clarity of understanding and that data literacy was really key to kind of helping the market move with us and understand the benefits of the changes we were making.
Juan de Castro: And what's the main challenge when some of the syndicates were pushing back? Was it the information that they had to provide you with was not available? Or you were asking that information in a format that was not straightforward for them to put together?
Cassandra Vukorep: The latter. Most of what we do, so reserving oversight, a syndicate or a manager, they will have reserving oversight. Underwriting oversight, they will have performance and overriding oversight. They'll just do it in a different way within their own company. And what we were asking them to do was manipulate what they do and care about with different basis points to interact with Lloyd's, which created a cottage industry of operations where teams only did Lloyd's reporting. So you're hiring FTE camps simply to create reports for Lloyd's.
Juan de Castro: So we touched very briefly on AI. This is one of the areas where AI is best, is transforming data. Shouldn't part of a Lloyd's vision be, okay, just provide me with the data in whatever format you've got it, and Lloyd's will take care of converting it to the format that you needed?
Cassandra Vukorep: That is the vision. So yeah, you've nailed it. And that's why we've started going out and talking to some of the syndicates about how they store their data, the basis that they have their data on, like what do they have? Can we just have that? And then we can manipulate it for our own purposes. There is basically a bit of a baseline, right? Because we're talking a hundred plus different companies. So there's still a bit of translation of understanding what they meant. And this is where data standards will come in and I'll harp on data standards and everybody needs to get onto the data standard. But the class of business and some of these elements, if we were speaking the same language within each one of these syndicates, then it would be really easy, even though they might cut and dice it in a different way. If we had a standard, we'd have a Rosetta Stone to basically work against to go, okay, well, this is how managing agent A defines how their class of business and how they do this. And this is how B does, etcetera, because we have it all mapped against. So it's really easy to bring it in. Because Lloyd's has to aggregate everything up. So we have to take everything together, aggregate it up for an insight across all hundred plus. It's getting a hundred plus different firms to speak the same language. And that's the challenge. And that's why the data council, the Blueprint 2 started. And the Data Council is not a Lloyd's initiative. Data Council is run by the market. It has brokers in it, it has managing agents in it, it has some of the key vendours that operate the market in it, but it is being designed to hopefully, for the entire market to be able to work better together. And that's where the need for standards kind of comes in. And if you talk to Microsoft, Google, like, yes, AI can do a whole bunch of things, but they do say, if a standard was there, hallucinations are less, it's better grounded, you get better analytics, it works a lot better. So there is an enablement with standards and AI.
Juan de Castro: And when you say standard, you can look at standards in different ways. One is a common data dictionary. So like when we say the data, the point X means something we all agree on the definition. And then you can take the standards all the way to, and this is the format in which we need to receive the input. So where do you think is the right level of standards?
Cassandra Vukorep: So what we're working on, so the CDR, so the Core Data Record for open market, and we've just finished Treaty. It's just come out of consultation with the market and we will launch this. And we do it in partnership with The Core, which is the largest insurance standards body. It has your data dictionary and it does have the format. What we are looking at, though, is not trying to create for every nuance and make it this big, overwhelming thing. Because, as you know, if you ask people for 20 data points, they'll work really hard to give you 20 data points really well. They're like, that's an easy ask. You ask them for 220, all of a sudden the quality will lax towards the end of it if they have to do, if it's not just a one-to-one map. And it's not a one-to-one map with a lot of it. So what we're trying to do, and that's why it's called The Core, is build The Core for operations. What is necessity that if it has a definition and this is the format and this is how we receive it and this is how the message would look, it would actually, instantly alleviate a lot of problems. And ACORD through Ruschlikon has a lot of examples of how, because of this, they've taken cash straight off of balance sheets. They've improved claims processing by over 40 to 50%. There's a lot of use cases that just simply having a good, solid Core instantly brings value to a business.
Juan de Castro: So touching on the CDR and the Core data record, there's a bit of a reflection, like what's working well, what's not, what's the level of adoption?
Cassandra Vukorep: Well, with the coordinated record where it's sitting at the moment is we do have kind of the Ruschlikon community who have adopted standards really going out and starting to use it between themselves and the brokers. So managing agents and brokers are both adopting it to go between themselves. Because Lloyd's has a Bureau in the middle, and that's why it had to be specifically created for us, until the Bureau is ready to ingest the message, it kind of limits the full impact of what that standard can have in terms of driving the ecosystem. So there's the benefit of adoption now, but the real benefit will come when we're ready to ingest it into the central services.
Juan de Castro: When you say Bureau, is that the same as the gateway? Or is it something different?
Cassandra Vukorep: No, there's only one in the Lloyd's market. But if we're talking standards, and this is anybody who comes in, anybody who's been in this market for 30, 40 years, will know that entity as Exchanging or the JV or the Bureau. And recently I had to explain to someone who came from a different part of insurance, life and pension. So like, but what's the Bureau? I'm like, no, it's all the same thing. They're like, what? I'm like, five names, same thing.
Juan de Castro: Which is like making it complicated.
Cassandra Vukorep: Yes, overly complicated. Because you think you're talking about one thing, then you're like, what? But what's this other entity? I'm like, no, same thing.
Juan de Castro: So then moving along to the topic of AI, I guess two questions. One is, where is the industry at right now in using AI? And let's start there. First question, where are we using AI?
Cassandra Vukorep: So varying degree of maturity, a lot of managing agents I talk to, they have co-pilot or some ChatGPT form in front. But if I ask, okay, but how much is actually into your process or enterprise level? When stacking, I can't remember if I'm stacking talking about that the other day, there's about 60% of the industry that is tried AI. They've got POCs, but only 8% have put that AI use case across five or more business functions. So it's not hitting that enterprise level yet. And that's the fundamental change. Like once you start putting it into your BAU business, that's where your real change will come from. You've got differing degrees. You've got the KI and the Hiscox who are at that level. And then you have quite a bit in the middle who are POCs thinking around. And then you have people who haven't started. So it's a wide range.
Juan de Castro: And what's preventing further adoption, especially, I mean, there are a number of case studies and successful implementations. What is hindering some of those going from either zero or just anecdotal usage as a co-pilot to properly embedding it in the business processes?
Cassandra Vukorep: I think it's the same problem we always have. People, process, tech. So how do you get off the legacy of remapping and redesigning your processes? It's the same problem that RPA had or machine learning had. Why are you just putting AI on bad processes? And then it needs that lean forward to go, well, actually, how do we redesign this process so AI can really do a bang for its buck? Or are we just going to automate processes that have been lying around for the last 20, 30 years and no one can tell us why actually we fill out six different forms for validation? Like whoever asked for those and who actually reads them? And to do it, true, you have to have a mindset of change. And it's that fundamental change in your business. And it's the reason we companies struggle with legacy architecture. It's the reason they struggle with process. Because it's hard. That's the hard work. But the companies that focus on it and actually get it into their processes and make it part of their ecosystem, their day-to-day working, not just helping them build an Excel spreadsheet or a PowerPoint, they will fundamentally in a couple of years be ahead of all their peers in the industry because everybody else won't be able to keep track with what AI will do for them.
Juan de Castro: So acknowledging the challenges doesn't sound like there's an option. So you're saying in a few years, unless you are really operationalising an AI in your processes, you will just be behind your competitors, right?
Cassandra Vukorep: Yeah. Well, you and I have discussed this. I have this theory. Could be a crackpot theory. But we're going to have to wait for one big guy to go. So it's someone who pushes a lot of business into the market. But when they are actually truly AI enabled, FOMO will kick in and others will quickly get on board. But some of them will be so far behind that they won't be able to get on board and they'll be left.
Juan de Castro: And is your hypothesis that would be a carrier? Or a broker?
Cassandra Vukorep: I think it could be either. I'm not as familiar with how, in true transparency, of how advanced brokers are in their AI evolution. But for us as Lloyd's, our direct relationship is with the syndicates and the managing agents.
Juan de Castro: And as you said, when a large entity, either a broker or insurer, really makes a difference. You mentioned FOMO, but is it FOMO? Or would you say, is it more when their peers realise that they are writing less business or they're writing worse business? Or there's actually like a direct business underwriting implication of not making enough progress?
Cassandra Vukorep: Yeah. If you look at it, the broker wants to do best for its clients. So they're going to go to the managing agent who can turn around something faster with accuracy when it comes to claims. It's not a whole bunch of fighting, you know, can actually be put out. The claim can be paid because the underwriting was done correctly. It was all within the terms and conditions. The faster that you can do that and more accurately you can do that, you're just going to put yourself on the front foot. And a broker is going to come to you because that's how their client is served best. So you will slowly see commercial value drip away because you're not as efficient.
Juan de Castro: That makes sense. And is there a role for Lloyd's in supporting the laggers in AI adoption, get up to speed or not?
Cassandra Vukorep: Again, it's not in Lloyd's job to devise the strategy of how companies manage themselves. We do oversight to say, can you write on our capital? Yes or no. But if you want to have AI or RPA or go blockchain, like that's really not in our capacity to opine on in either way. We obviously want the market to be healthy and we want the syndicates that are running business to be healthy, vibrant entities. And if adopting AI is one of the ways that they need to get there, then yes, we would like them to kind of look at it. But that's not something that we in any way make a decision on. So internally, which we will roll out, we have a data community of practice, which AI is a big part of that. And our data community of practice is not just data people, because obviously data people love talking about data all the time. So we don't really need to convince ourselves of these things. But it brings in people from markets, underwriting, finance, asset management, etcetera, to kind of learn about data governance, why data ownership is really a part of the foundations and the fundamentals of data literacy, and then the enablement of AI in there. And so we've got people doing prop training, like how you ask the right question. I had to go through it because I am one of these people who is so old where I just assume everybody around me has the context. So I say three words and it gets done. And I had to remember. No, no. Remember when you used to work for juniors, you just came out of university. Like you had to spell out everything in a lot more detail. That's what your prompt needs to look like. So if they can teach me, they can teach anybody. And those are some of the- And how we're trying to get around. Like I said, we haven't paid for Copilot, but we have most people in the market are on Microsoft. So you have the Copilot that's part of the Bing Chat, the Microsoft website, and it has enough power to do quite a few little things that will just make your life a little bit easier. So the recent one that we did for the corporation is objectives. I don't know about you and your company of how what objective time comes around, but it's a bit more of a drain than a then like, let's move forward. And then are they smart objectives? Are they aligned to the strategic pillars, etcetera? And you've got an entire corporation trying to align in one way. So the team simply just wrote a prompt that would do all the things and then where they put it in. So now the AI is screening. This is AI Studio. So not the cheap Copilot to see if the objectives across corporations are smart, if they're aligned, etcetera. And from last year, this year, there's been a huge uptick. Yes, the percentage is much higher in alignment of the culture and the corporation pulling together. And it was simply one prompt being written and then everybody using it. So there are little things that you can do that won't cost you an arm and a leg.
Juan de Castro: So, talking about objectives, what are your objectives for this year as a CDO?
Cassandra Vukorep: So my main objective for this year is reducing the burden on the market when it comes to the operational complexity that we put out there. Because the majority of what comes in and operates through Lloyd's is data. We ingest it and go through a washing machine and analytics and stuff, and then we spit it back out. So it's fixing that so it's easy for the managing agents to give us the information. And it's very streamlined inside for the teams to do the job that they need to do to oversee the market. And that's it. And that's my big objective. It's just small.
Juan de Castro: You can say it in 20 seconds. It's huge. I mean, if you achieve that by itself, right, just reducing the burden while improving the quality of the data and being able to aggregate it across the market would be obviously a huge achievement. And if you had to say, like, let's imagine you're in the role, in your current role in 2028, just a little bit like three years away. Any predictions of, like, what you would want to have achieved in those three years that would make you proud of your current role?
Cassandra Vukorep: Couple of things. Number one, we're no longer talking about market reports. They're APIs or data feeds that come in, and actually the client that the syndicates don't even know it's happening. It's just so passe. It's just something that they need to do, but it's so simple. On the Lloyd's website, actually a dictionary of everything we ask for and why, to bring clarity and transparency. And the last one is the expansion of the data community of practice of where it's not just run by Lloyd's, but it's living and breathing from the market. And they've taken ownership of trying to share data, data governance and data literacy amongst ourselves for the viability of the market itself. And it doesn't actually need Lloyd's to drive it. Those are my three things. Obviously, the data standards have been fully embedded. Missed it.
Juan de Castro: Yeah, way before then, that's for sure. Perhaps we should then schedule another podcast episode in 2028 and look back because I think we would be surprised. I'm sure you will make much faster progress than we think today in many of those areas. But at the same time, I'm sure there will be some others where we would have thought that we would be farther ahead and we're still struggling with some of it. So it would be interesting, like, in what areas have moved faster than expected and which ones have moved slower than expected.
Cassandra Vukorep: I'm interested to know in three years where the technology has taken us and what we're actually talking about. I think it's the one thing when I'm building a strategy, you always get down to the tech and people are, what are you picking? What are you doing? Analytical tools? Is it business objects? What are you doing? And my answer always is, I have no idea in two years what the best in breed is going to be. All you need to do is build me an architecture that is flexible enough that for me to do a migration from tool to tool does not take three years. It takes six months and it doesn't take six million pounds. It takes 500. And that's all I can strategically say because the pace in which AI is moving. I don't have a crystal ball to tell you where we will be.
Juan de Castro: Well, we will check in a few years time and look back. Cassandra, it's been absolutely fantastic just to get a sneak peek at your role as CD and AI officer at Lloyd's, learn a bit more about your priorities and objectives. So really appreciate you joining me today.
Cassandra Vukorep: Thank you very much. Appreciate it.
