What is AI delivering so far

Record date: 9/25/25
Air date: 10/22/25

Generative AI is transforming business at an unprecedented pace, but are organizations seeing real value from this rapid adoption? In the debut episode of Zurich North America’s Future of Risk AI miniseries, host Justin Hicks speaks with Kristen Bessette, Chief Data Officer at Zurich, and Anthony Meyers, Managing Director of Data and AI Strategy at Aon, about the practical benefits and pitfalls of integrating artificial intelligence into business.

The conversation highlights that while AI adoption is accelerating across industries, success relies on understanding its limitations, aligning technology with business strategy, and embracing change. Real-world examples show how AI is already streamlining workflows, enhancing decision-making and improving customer interactions, but achieving true return on investment requires incremental progress, strong governance and ongoing learning.

Business leaders are encouraged to start small, experiment and include cross-functional teams in their AI journey. Ultimately, AI is not a magic wand—but when implemented thoughtfully, it can empower organizations to navigate disruption and unlock new opportunities for growth.

In this miniseries, other episodes include:

11/5/25: Five ways everyone can benefit from AI today
11/19/25: Dark side of AI
12/3/25: What’s next in AI?

Guests:

Kristen Bessette
Chief Data Officer
Zurich North America
Connect on LinkedIn

Kristen is Chief Data Officer for Zurich North America, where she is advancing the company’s data and artificial intelligence initiatives to optimize performance across core insurance functions. Before joining Zurich North America, she was the Chief Actuary, Data and Analytics Officer at QBE North America and earlier held senior leadership roles at Liberty Mutual Insurance. Bessette, who has become a leading industry voice on AI in insurance, is responsible for all aspects of data management for Zurich North America as well as the development of advanced analytics and AI capabilities and products. Over her career, Bessette has worked on growing and rehabilitating parts of businesses, using her quantitative expertise to evaluate risk and pricing and more broadly to use data and analytics to achieve better results. Bessette is a Fellow of the Casualty Actuarial Society and a member of the American Academy of Actuaries. She holds a Bachelor of Science in Mathematics from Bates College.

Anthony Meyers
Managing Director of Data & AI Strategy
Aon
Connect on LinkedIn

Anthony is a Managing Director at Aon, focused on data and AI strategy across insurance, reinsurance and broader business. He works with leaders to turn AI and analytics into better decisions, deeper insights and actionable outcomes. His work ranges from enterprise AI adoption and development to risk mitigation and risk transfer optimization, with measurable business results at the core. He also speaks fluent buzzword, including big data, machine learning, AI and emerging tech, but prefers plain language and outcomes.

Host:

Justin Hicks
Communications Business Partner
Zurich North America
Connect on LinkedIn

Justin Hicks is a Communications Business Partner at Zurich North America and supports enterprise communications efforts for the Direct Markets business and the Operations and Technology function. Before joining Zurich, Hicks was the first dedicated internal communications manager at Rivian's electric vehicle manufacturing plant in Normal, Ill. Earlier he served as a public affairs communications specialist at State Farm, supporting claims executives and leaders.

(PLEASE NOTE: This is an edited podcast transcript, capturing speakers with natural speech patterns that may include incomplete sentences and/or asides, grammatical errors, verbal shorthand and some statements that may be less clear in print.)

EPISODE TRANSCRIPT:

JUSTIN HICKS:

GenAI is considered by many to be the most rapidly adopted technology in the history of the world, but is it delivering real value? Welcome to Future of Risk presented by Zurich North America. We explore the changing risk and resilience landscape and share insights on the challenges that businesses face to help you meet tomorrow prepared.

We're kicking off our AI and machine learning podcast miniseries by looking at what meaningful benefits AI has created so far. I'm your host Justin Hicks, and today I'm speaking with Kristen Bessette, Chief Data Officer at Zurich North America, and Anthony Meyers, Managing Director, Data and AI Strategy, at Aon. Kristen and Anthony, welcome to the podcast. How are you?

KRISTEN BESSETTE:

Great. Thanks, Justin.

ANTHONY MEYERS:

Yeah, very well, thanks. Thanks for having me.

HICKS:

Fantastic. Anthony, it's great to meet you. Kristen, always a pleasure.

AI adoption: Between hype and real value

HICKS:

I wanted to start with each of you and ask, is there a pet peeve or is there a hot take of some kind that is really irking you or driving you crazy about the dialogue around AI

MEYERS:

Happy to jump in here, Justin. I think AI is, it’s an interesting topic, right? Because it is probably one of those technologies and hurdles that we’ve cleared that is going to be human species-defining — which is a bit philosophical, right? — beyond typical insurance and financial services conversation. The question that I ask people is: “to do what?” I think we hear often like, "Hey, AI, AI, AI." To do what? And there's sort of two camps that I feel like I spin when I talk to teams and companies.

The first camp is: It's not perfect, therefore we can't use it, right? Again, in the industries that we operate in — highly regulated, require a lot of compliance, require a lot of risks and security measures — there's that camp.

And then there's the other camp of “We have to use it, now! When are we doing it? What are we doing? Is it working? Where is it going?”

And there’s sort of this happy medium that I think we need to find, that we’re starting to see come up in just like the general usage trends, right, with some of the big gen AI players putting out their reports over the last couple of weeks. And I think some eyebrows are being raised at the usage figures in, again, both directions, right? I think for some people like myself who just have that expectation that everybody uses AI as much as I do and sort of everything that I use it in, [the usage seems] probably pretty low. But I think for the general people, general use cases, [usage seems] probably pretty high, right? I think the stat was something like 40% of the U.S. population between [ages] 18 and 64 is using it, right? So, there’s sort of a happy medium. Again, it's not lost on me that the rate of adoption of this technology is probably faster than any technology we’ve ever seen on the face of the earth. So, I think one of my many pet peeves is, let’s try to find the happy medium with it, we can use it safely, it is driving value — but let’s not think it’s the silver bullet that everybody is kind of hoping it will be.

BESSETTE:

I would just echo what Tony said. I think, you know, my GPS has taken me to the wrong place before and I didn’t throw it away. I still use it. I just understand when I pull out of a parking garage in Manhattan, it’s not going to pick up where I am right away, and I’m going to have to know where to go until it finds me and gets me on my route, right? So, I think you have to understand the limitations. You have to understand where the models are strong, where the models are weak, how it applies to what you’re doing, and also understand that AI isn’t the answer to every problem that you have.

You know, a lot of times people come in and say, “I want AI to do this,” and we have to ask, “Well what is your problem?” Because there might be a more straightforward solution, a cheaper, faster way to get you there, or at least 80% of the way there, that doesn't require a big buildout of an AI model. If you have a fly, you don’t have to sledgehammer it, you can just swat it sometimes, right? So, I think if people are thinking about that and come to their partners with problems instead of, “I need AI,” you can get to the answer a little bit more quickly, too.

HICKS:

And I think both of those comments are well said. But here’s where I get a little vexed, I guess, in my own thinking about it. So, we talk about it on this grand scale, you know what I mean? Anthony, in our pre-meeting we even talked about how this is one of the greatest innovations in human history. And I think about all of these different innovations going all the way back to the discovery of fire <laugh> and the wheel and movable type and radio and trains and automobiles, and you know, television and the internet. And now here we are, with this grand innovation at our fingertips that we know has the potential to change life as we know it or so would seem. But then you say it’s not a silver bullet. And I think people look at it and say like, well, why is it not delivering all of the immediate good and cutting out all of the bad? Why is there not any sort of gains that I can see with immediacy? That just prevents us from cutting out all the negativity around it. I think there’s still some issue there.

BESSETTE:

And I might jump in, Justin.

HICKS:

Yeah.

BESSETTE:

And just say, like, you mentioned trains, right? Trains were transformative. And if you’re a horse-and-buggy person and didn’t get to the train, you’re out of business, right? But there’s a period where you’re laying tracks through the desert where you’re just putting cost in and you’re not seeing a lot of reward. But when you connect the coasts, the world is different, right?

HICKS:

Right.

BESSETTE:

And so that’s kind of — a lot of companies are sort of laying tracks on their AI journey. They’re doing small little spurs out to places, they’re getting little wins, but they aren’t seeing that kind of transformational thing actually hit their ROI yet, because we’re still in the investing-and-exploring phase, right? That doesn’t mean it won’t happen, but I think there’s work to do for a lot of places, and people need to adopt as well, right? People need to learn how to use it. People need to start exploring, people need to start changing their processes to actually work with AI and embrace AI and bring that technology to their jobs.

MEYERS:

I think to further extend some analogies; the runway is being built as the plane is taking off. Listeners need to recognize that 10 years ago, generative adversarial models were just kind of starting too, you know, I can appreciate it’s more than, come into popularity. And then in 2017, the transformer architecture came around. So, for better or worse, we’ve lived through a lot of history over the last two decades or so. And 20 years from now, we’re going to look back and go, “Oh, yeah, like, that was super transformative.” It’s getting better by the day. So, what maybe wasn’t providing value yesterday could provide value tomorrow. And so, as we think about this from an adoption perspective, right? — one of the things Kristen, I think, commented on is, as we think about process, right, we’ve been so used to dynamic, or sorry, static, like parallel-type processes. So, if we take underwriting, like you get the data, your risk score, it goes to somebody else, it goes to somebody else, it goes to somebody else. It’s this really nice straight line that it usually happens in and maybe there’s somewhere around the line that it doubles back and somebody double-checks something. So, we’re used to this sort of linear thinking where, what AI allows us to do is, you know, hey, we should probably sit down and think like, if we can web this out, so, you know, think of a multi-pronged tree that starts to happen, all happening at the same time, like, what does that allow us to do? And so, you know, part of the thing I think that people and corporations and enterprises are struggling with realizing value on is the rethinking, like, how do we think about this? So, you know, using GPT models is not the same as using something like a web search. [It’s] learning how to use GPT models and also just appreciating how fast it’s changing as I was just commenting on, right? And the question that you led with, the first thing that we had to do was figure out how to get rocks to process electricity properly, right? The computer chip. To then like designing computers to then building massive cloud infrastructure to then building massive electricity-producing facilities to then figuring out how to get binary integers to compute properly, and then at scale. And then you’ve got GPUs.

And now all of a sudden, we’re at this point of like, “Oh, hey, I can write an email like a pirate. That’s pretty neat.” But how do I realize business value on that? And again November of 2024, 2023, ChatGPT rolls out, everybody says, "Hey, we need to use this." Maybe we realize quickly, “Look, insurance and finance [are] a little more complicated than just drafting some messages.” I think now we’re starting to hit a rhythm. We're starting to identify the value creation opportunities, but it’s not the universal approximator, right, that sort of, everybody’s hoping it will be.

BESSETTE:

And there’s some speed bumps, right? You know, is your data set up to do this? Do you have the right infrastructure to do it? Do you have these things, right? That will slow you down until you level those out. So, I don't know if we’re done with our transportation analogies. We may roll ’em through the whole time, but --

HICKS:

Why not?

BESSETTE:

But it is true, right?

Using AI to drive business strategy

HICKS:

I think they're working. I think they're working. Yeah. No, and I was going to say, that Anthony, your comment, I think was the perfect segue to the next kind of iteration of the conversation being that we hear all about the potential of AI. So, what are the actual ways that your organizations are using AI to drive strategy at this point?

MEYERS:

So, we’re really focused on it like on an everywhere basis, right? So, sports analogies, you know, Minnesota Vikings actually doing all right this year, defense is OK, right? And so, we think about like covering the field or flooding the zone, whatever terms people want to use. We are looking at it from both a top down, right across the enterprise, what are things that we can scale rather quickly? Things like our internal GPT model, AonGPT, things embedding into our core systems, right? So, Kristen made a great point, is our data ready? Are our data systems ready? Are we thinking about data quality? Are we thinking about the insights that the AI is going to drive? And also, a bottom-up approach. Like, there is just a reality that we have a lot of knowledge, training and expertise in the industry that our colleagues can provide us, and we want to give them the tools to be able to drive innovation in the domains that they know.

One of the things I tell teams often is, look, I can build the biggest, sexiest AI model, AI tool that this company’s ever seen. You'll beat it every time just by having 20 plus years of domain knowledge. Like, you understand your business, you may have built the business, right? The reality is, certainly within Aon, there are people that have literally built the markets, like they understand it, like A to Z. And so, we’re seeing a lot of value, I think, in what I’ll call information retrieval, right? So, sort of standard RAG patterns, if you will, upload documentation. You know, definitely with sophisticated actuarial, analytics, right? Modeling type of documentation that might be several hundred pages long, and you’ve got to figure out, you know, why something is the way it is. One of the other things we like to say in Aon is the insurance industry, if it’s not good at anything, it's definitely good at creating PDFs and forms, right? So, contracts, slips, quotes, right? Things like that, binders, and just being able to extract information from that. So, extract information to help identify trends. We’re really good at tabular data trends. What about unstructured data from text, video, sound, podcasts? And then, you know, I would say code generation is a big one. AI-assisted coding. We’ve seen a lot of value in — caveat that with, it’s not perfect code, but it’s certainly great to get you started. And then making data more accessible, I think is one that we’re really starting to see the value in. So, things like text-to-SQL or any of those type of capabilities that are really democratizing the ability to interrogate data versus needing to train a whole bunch of people how to write SQL, Python, R-type coding to just get relatively simple answers.

BESSETTE:

And if you think about what that means to any of the employees, right? You have underwriters that are interacting with third parties, they have this whole process of exchanging documents, all of those capabilities that were just mentioned, to allow you to extract the data, pre-fill it, augment it from third parties, summarize it, start some communications, all of those things can happen. Claims intake happens faster with that same sort of thing. So, you end up with a more frictionless environment, both for your employees, your trading partners, your customers to be able to interact with you. And you can produce — with all of that interaction with the data — better insights, better things, better information comes out faster so that your business can be more successful. So, you can take your business strategies and have better topline growth, better profitability, and so really power your business strategy through AI and other technology to actually get the results that you have in a better way.

MEYERS:

Kristen, I think you hit two major points. Let’s say AI never provides any value from a technology perspective. Let’s just assume it’s not. What if one of the major benefits is just letting people do what they want to do more efficiently? Like, we reduce that friction, internal friction that I know we’ve all experienced, whether it’s just getting data for someone, or like this file says, you know, “final V three actual do not use,” like, is this the actual one or is there a different one, right? Like, just getting that relatively simple stuff to do what you need to do, but then also like, “Hey, what if we could improve work life balance?” That’s a huge value prop of some of the AI techniques, and I just wanted to call those out. I think those are excellent points.

Enhancing business outcomes with seamless AI integration

HICKS:

What I'm getting out of this is it saves money, it saves time, and it provides simple answers faster. Is that bottom line enough for you guys? Or is it, is it deeper than that?

BESSETTE:

I think that’s the potential, right? I mean, right now I think people might argue that you’re not saving money because a lot of people are investing to get to AI solutions, right? But I do think that’s the potential that we see is that you can look at this and get better business results because you’re better able to make decisions faster, more seamlessly and provide just better interactions with all your trading partners, which should lead to people having more capacities to actually serve their customers, right? Because you’re not in these manual tasks like we just talked about. You can actually go to the more complex tasks that need understanding of the market, that need understanding the risk profiles of the customer, that need somebody to help solve problems, because your day isn’t taken up with the mundane as much.

HICKS:

I mean, I think most people would sign up for that, Anthony, right?

MEYERS:

Yeah. Annoyingly I use a phrase often within Aon, which is, "it depends," right? So, I think Kristen hit a bunch of the points. I think one of the things that I would encourage people, listeners, to be prepared for is sometimes when you pick up the rocks, you find things you didn’t even know that you were missing, right? And so, as we think about the ability of AI to, you know, rip through data, provide insights and whatnot, sometimes we find things that it's like, “oh, we've been doing this for 40 years? We’ve got to remedy this pretty quick.” And so, I certainly think the opportunity is there, especially around things like topic research, company research, right? Things that would otherwise have taken more time than probably was generated in value now become relatively novel exercises.

Imagine being able to query your business data daily. How many large losses did we have today? What’s the loss ratio today? Again, fair warning, sometimes you may not want to know, with that much velocity, but the ability for people to just make business decisions, again, based on their knowledge, training and expertise, because they have access to this data, beyond some of the value of the automation, which we've been doing for a while with, with RPA and things like that, I do think has a lot of potential, not only in our industry, but again, across everything — education, healthcare, really every industry.

HICKS:

And not only that, but even in our personal lives, I know that I’ve seen the benefits of it myself. I do a podcast on my own, outside of Zurich, and I do research for that. And I remember I recently used ChatGPT to compile some data that I was doing to make a point on my podcast. And the research was gathered instantaneously. And I’m talking like, what would’ve taken me several days to gather, I mean, it just spit it back out in a matter of seconds. And so, it was stunning to me how fast that all came about. Also, we’re shopping for a new car in our household. So, it’s like, “hey, what are some of the specifications of the type of vehicles that we’re looking for? What are some vehicles that are available in the area?”

The unprecedented pace of AI adoption across industries

HICKS:

And it's not perfect, you know what I mean? But the groundwork is there. And so, the potential is really exciting. You guys talked earlier about pace of adoption. And we even talked at the top of this show about the rapidness of the adoption, that this is the most widely adopted or most quickly adopted technology that the world has ever known. That's a big statement, right? How would you characterize the pace of adoption then, both inside and outside the insurance industry? Because I think it may vary a little bit.

BESSETTE:

The pace in both, I think, is faster than we’ve ever seen. I don't think they're equal for insurance and retail, but it's still faster than we've ever seen, if you look at the broader statistics. Like you said, Justin, in your personal life or just any number of companies in financial services or outside are really just exploding with the use of AI. You know, insurance does have some unique challenges. It's a complex product. It's a regulated product. There is, you know, more risk around giving the wrong direction potentially than in, like your car shopping example, if they told you that a black car was there and it turned out to be gray, it's not maybe what you want, but it's not a terrible outcome, right? You can go somewhere else and look for it. So, I think there's different trade-offs that we have to make, but I think in all cases, the pace of adoption is extraordinary around the world and in all facets of life, really.

Overcoming AI fear and embracing opportunity

HICKS:

And I think one thing you also mentioned in our meeting beforehand was that the benefits are outweighing the fear, is that kind of what you've observed?

BESSETTE:

Yeah. I think the biggest risk is not doing it, right, is getting left behind, kind of back to our trains, right? If you're not on the train, you're out of luck when the train is moving, right? So, I think there are a lot of great things that are going to come out of this. You need to be, if you just think back to your individual work experience, you know, take the opportunity to learn this, right? See how this affects your job. See how you can lead the change for your organization because a lot of people are worried that their jobs will be replaced, that they'll lose their jobs. But there are people in the workforce today that didn't have computers on their desk when they started working, right? Did they have to learn new things? Yes.

HICKS:

Absolutely.

BESSETTE:

Is their job the same as it was 40 years ago? No. But they still have jobs today because they did those things that they needed to be successful in that changing environment, right? So, if you're worried about it, you know, be the person in your team that drives that change. Be the person that adopts. Understand how AI can make you more powerful as an employee and embrace what's happening. And I think that will make a big difference.

HICKS:

Yeah. Not to go down a road here, but is the fear different now? Because like you said, people have had to innovate in their roles before, so why does it feel like it's different now? Maybe it isn't. I don't know.

BESSETTE:

I think there's a lot of people talking about it, right? So, talking about how everybody won't have jobs and it'll be in places where people have not been affected by it [previously]. There's certainly been jobs that are a lot different if you look at the impact of robotics in manufacturing or some of these things, right? You know, people definitely are not employed in the same ways in those industries as they were 30 years ago. That is true. That doesn't mean that there are no jobs anywhere for people, that we just have this huge amount of unemployed people drifting, right, there. You know, there are a lot of opportunities created by AI. There's also a large talent gap. If you look at the demographics, people have been talking about, “Hey, we’re going to have all these talent gaps.” Maybe we need this type of productivity boost to actually fill in some of those gaps, right?

BESSETTE:

So, I think there's opportunities. I think people need to go out and seek them and learn and embrace the new technologies like they always have. And accept that the world is going to change at a faster pace now than it ever has before. And that part, I think, is truly different. The pace of change here is extraordinary. Things that we thought were cool, that we were really excited about building one or two years ago are now commodities that we would just buy from any number of people, right? So, that pace of change is really extraordinary now. And that piece will continue because AI is going to fuel the pace of change. Now, it's going to be not as linear. It's going to be a lot more exponential than it ever has been.

MEYERS:

It was interesting, Justin, just this morning -- we're recording here this afternoon -- we identified another job function that AI was going to create, right? Because if we think about all these RAG [Retrieval-augmented generation] patterns and the knowledge management, like who's managing the knowledge base. So, somebody will come up with a good term, “knowledge manager” is maybe just another good standard one for our industry. But, you know, Kristen had a great point there, right, it's not like people that are trying to do really, really futuristic thinking things. Again, quantum computing is a big one. Space travel, right? The future of mobility, the future of healthcare, they're not going to slow down because of AI, right? AI is going to help them move faster, right? So, within our industry are we prepared to underwrite space travel?

Are we prepared to underwrite, you know, let's talk superstar Trekkie-like teleportation of goods, because we figure out how to quantumly teleport things, right? What does that do to the shipping industry? How do we have people sign off on it? And so, getting comfortable with, you know, it's the adage, right — getting comfortable with being uncomfortable — I think is going to be a requirement. But you know, the thing that's inspiring that we talked about in our pre-meeting as well. If we think about the speed in the healthcare industry at which we can sequence viruses and come up with vaccines, or what's happening in oncology, right? The ability to identify cancer from scans, it's just unbelievable. And the thing that I try to encourage people to do is that AI can teach you how to use the AI as much as we can learn things from the AI. So, if you're sitting here wondering like, how am I ever going to use AI? Ask the AI, right? Like, “Hey, this is my job function” — you know, be mindful of your company's privacy and security considerations; don't go just throwing your data up wherever — but “Hey, this is my role, like how could I start to embed AI into my workflows?” It'll give you a nice little response, and then maybe try some.

HICKS:

I love it. I love that. And one other thing that you pointed out previously was that AI is not an IT tool, right? And I think we're clearly starting to see that it has the capacity to, or if in fact it will be a part of all of our business in every way. It's not something that is strictly relegated to, oh, that's an IT thing and I don't do it, therefore I don't have to concern myself with it. No, it has the possibility to supercharge your work no matter what field you're in or what industry.

BESSETTE:

And more people are using it today than you think. They might not be using it at work, but they're using it on their phone. They’re using it in their personal lives. So, take that curiosity, take that piece and bring it into your work. Like figure out how it works and embrace it.

MEYERS:

Yeah. I would say at Aon we look for very pragmatic applications of AI. And how you find pragmatic applications is by understanding what the business strategy is. You know, the analogy of the computer, right? The computer comes around, it's a technology team problem. It's an IT team problem. And I love my technology and IT teams, right? Love 'em to death, but I don't think when people, when the digital operating model first started to come around and we'll avoid some names but I’m sure people can think of them, right? It wasn't the IT team that was driving the new digital frontier of business, it was business teams. And the reality is everybody's a part of the business team because everybody operates within a business. And so, I would encourage teams, people, enterprises, corporations, sports teams, right, [to] lead with the strategy and then let the technology help that instead of trying to fit the technology into the strategy. Because you might find that your strategy needs to change.

Getting started with AI: Practical steps for business leaders

HICKS:

And businesses have to be nimble. We certainly know that you have to be nimble in order to survive and thrive ultimately. So, we've covered a lot. And I'm curious, what's one piece of advice that you would give a business leader who is either trying to adopt AI or adapt to it, and how their stakeholders can help navigate this revolution.

BESSETTE:

I mean, I think it is just, “Get started,” like Tony was saying. You can get started very easily by just trying. Go in and interact. Most companies have an internal system. So, use your internal system, to your point [Tony], stay within your walls to do these things, but go in and start interacting it. You know, a lot of the tooling will be embedded in your Microsoft products. You might have access to Copilot natively. You can use it to summarize your email after you've been out on vacation, right? You can use it to do any number of things that would just allow you to see some of the things that make your life easier. So, just start trying things and have your team start trying things and just sharing the things they've done. And you'll get ideas from all of that. And then you can look at that. Those little incremental things can actually add up to big changes.

MEYERS:

Yeah, that's a great one. I think the one that I push often is, “Read.” Right? There's a lot of stuff out there. You’ve got to read and read and read and read and read. And again, the nice part is you can kind of let the AI help you summarize, like, “Hey, what happened today? What happened yesterday?” And so, staying up to date is going to be key. You know, Justin, this is actually one of the key AI risks that I see in business. It's one I talk about often, and it's not a technology problem. It's a “do you understand AI enough to know what it is and isn't capable of,” right? So again, this dream future of everything being automated and everything being straight-through process and like, there's no friction anywhere and blah, you know, all of that. Are we ever going to get there? Maybe.

But we're not going to get there if we also have to be nimble and have to adhere to, you know, regulation and compliance and security and privacy and things like that. And so, understanding what it can and, importantly, can't do, understanding where it is likely to legitimately disrupt your business operations, your business strategy is going to be vital. And so, you know, for those of you sort of mid-career, thinking, what's going to happen in 20 years, maybe in five years, right? Like little bit of design thinking. If you haven't heard of design thinking, go look it up. It extends to everything, right? You know, decompose your business. Think of your business model. Take a pen and a piece of paper and draw it out. Where do you think AI can be applied?

Again, information retrieval. Great. What does that mean? Better chatbots. OK, what does that mean? Well, it means I change how I interact with my customers. Well, then what does that mean, right? And you start to do this exercise of looking forward, but it's only possible if you can understand. And I can appreciate that it's a little out of one side of the mouth, out of side of the other, because who knows, right? AI space could change in two years, but as of right now, understand what it can do, understand where its limitations are, and think about how that applies to your business in as agile of a framework as you possibly can. You certainly don't want to build big monolithic software programs, hardware, infra-architecture, things like that. But just start learning. You’ve got to start learning it.

BESSETTE:

But you do want to do it about the totality of your business, like Tony just said. Because the more you say, I'm going to solve this problem in this spot and that problem in that spot and that problem, that spot, with AI, the less optimized your solution is across your whole business process. So, when you're thinking about where I could do that, lay out what you're trying to do, like you talked about, and then said, where can I improve this through the use of AI? And then bring those pieces in so you end up with a cohesive environment at the end, of solutions that actually make you better and drive your business forward. And I think that's a little bit of a different way than people might be thinking about it now in some applications, right? So, the more you can think holistically across what you're trying to do and what you're looking to do a few years ahead without building out solutions that can't be changed — because we all know everything's going to change very quickly — you know, the better off you'll be going forward.

MEYERS:

I think my second level of advice is, and I can appreciate that there's going to be a bunch of people that are about to roll their eyes or maybe even turn it off, right? —

HICKS:

Let's hope not. Let's hope not.

MEYERS:

— is you’ve got to lead with governance. Like the governance space in AI on the planet right now is, it's complicated, right? And so, you have to lead with governance, you have to have the frameworks in place, you have to have the processes in place. You know, what models are approved in this jurisdiction? What is a high-risk use case? What is in a high-risk use case? When and what can I or can't I use this specific data field? We probably spend more than half of our time when we talk about AI, talking about governance, compliance, privacy and security. If you don't do that and you build out all these use cases, you're going to have a bad day. So, set your foundation of governance and compliance and then build out your business strategy.

BESSETTE:

And to know what's high risk and low risk, you have to define your risk appetite. What is your appetite for something happening in this space? And like I said, that's going to be different for your car shopping example, Justin, than it is for a claim’s coverage decision. At least I think so. So, make sure you understand your organization's risk appetite, have the governance that's appropriate, and the procedures that are appropriate to operate within that risk appetite. And that's a really key part of execution broadly, as you're building out your AI strategy.

Why business agility is key to AI success

HICKS:

Anthony, there was one word that you said that stuck out to me. It was disrupt. Just the disruption of, I think, people's business may seem daunting. Disruption doesn't have to be a bad thing. In fact, I could argue that it's a good thing because it, like you said, it makes you uncomfortable. That's how you stretch yourself, and that's how ultimately you become better, not only internally for your employees, and not only as an organization, but ultimately for your customers, which is the end game. So, I think to the extent that we're able to do that, I think businesses are in a good position to move forward with AI and to explore and to play around with it and see what the possibilities are, because there really seems to be no lid on where this can take us. Any closing remarks that you guys would like to add to the conversation? It's been a really, really fun discussion today.

MEYERS:

Yeah, [crosstalk] go ahead. Go ahead.

BESSETTE:

I think it goes back to just kind of summing up the stuff we've talked about, right? There's terrific potential here. So much change, so much capability that's just outside our reach, I think. So, it's a really exciting time, I think, for companies and for employees to actually jump in and to start working with AI and to try to achieve some of the things that we can see, right? So just embrace the change and you know, buckle your seatbelt, sign up for the ride.

HICKS:

Embrace the change. Absolutely.

MEYERS:

Yeah, I think there's, I probably have too many, so I'll put three out here and,

HICKS:

We've got time, man. We've got time. Go for it.

MEYERS:

<laugh> And so, the first is, you know, I have this conversation often, is, back to our horse analogy, there's still a lot of rolling around in the mud that has to happen, right? Like, data isn't going to clean itself. Understanding the processes isn't going to do itself, right? The AI is very helpful at doing that, but we still need to do that, right? We still need to make sure we're getting the data where it needs to be, when it needs to be, so that it's accessible when we want it to be.

My other advice is don't overbuild. It's super tempting to sit down and try to do 18 months of planning on development for AI. I would encourage any developers or any people that vibe coding, which candidly is something I despise even though I do it a lot, instead of capability delivery and instead of user experience delivery schedules, which are still important, think about it from a value delivery perspective, right? If you can deliver value with AI in 30 days, do it and then capture feedback, measure it, another 30 days. And that's going to allow you to, instead of making, kind of the big shifts, if we think of software development, you get requirements, you build their requirements, they sign off, right? It allows you to manage that change, which is happening quickly. So, instead of like big sweeping motions, you should have really fast, jittery motions. And then I think the analogy which I have to wedge in here because I'm told it's really good, this may be North America, but anybody that follows soccer-football will get this, right? One of the things that the best midfielders that have ever played the game do very well is sometimes they stand still, right? They let the game move around them. They see what's happening and then they move in the direction they need to move. I would encourage many of you, if not all of you, to operate that way in the AI space, right? Chasing the shiny object is a great way to spend a lot of time, a lot of energy, and a lot of money, and not get very far. So, think about how it aligns to your business strategy. Remember that like good old-fashioned data cleaning, while it can be supported by AI, there's still going to be cases where you don't want some data to go into a model, don't overbuild and stick to your guns, stick to your business strategy.

BESSETTE:

Yeah. Because it really is, right? Technology, including AI, is there to empower your business. So, your business should lead the change and you're there to make that happen, right? And so how do you do that the most effective way? [I] completely agree with what Anthony said in terms of the way you think about that, particularly in this time where everything's changing.

MEYERS:

But you also can't do it alone, right? So, when you, when you're thinking about AI, have the business in the room, have the technology team in the room, have the data team in the room, have the analytics team in the room, have legal and compliance in the room. Because it, look, we're all learning this together and the industry is changing together, right? And so, to be really successful, you need everybody in the room.

What is vibe coding? Accelerating app development with AI

HICKS:

So, Anthony, I heard you mention “vibe coding” earlier. Can you explain what that is for our listeners?

MEYERS:

Yeah, so vibe coding is a relatively modern phenomenon. It's a workflow idea where you use AI to help generate code to very rapidly develop apps, software and the like. Super, super useful for prototyping. I think what many people listening will probably understand is enterprise scale is a lot harder than a simple app on your desktop. So again, a lot, a lot, a lot of value for AI-generated code to get started. But there's a whole world beyond that that, I just want to make sure people recognized.

BESSETTE:

And so, what I refer to it as [is] “show but not go” because it's great for demonstrating things, it's great for idea generation, it's great for getting a consensus on what you want things to look like and where you want to head. But you can't just flip the switch and roll it out to thousands of people. There's a lot of work that has to happen before you would actually put that into production.

HICKS:

Well, it's pretty obvious that AI is delivering a lot so far. And the possibilities again, are to the moon. It's just no limitations as I mentioned earlier. So, you guys have really done a great job encapsulating that, and I appreciate your time today. Thank you so much, Kristen. Thank you, Anthony, for joining us. And thank you all for listening. Stay tuned for our next episode in this series where we explore AI literacy and the many forms that can assume within your organization. Our guest will be Jay Bell, Head of Innovation for Specialties and Head of Operations for Construction at Zurich; and Mark Breading, Senior Partner at ReSource Pro. If you liked the show, leave a comment or review wherever you get your favorite podcast. Or you can drop us a note at media@zurichna.com. This has been Future of Risk presented by Zurich North America.