Faster, Smarter, Riskier? Inside the New Actuarial–Technology Frontier

In episode 1, Steve explores how actuarial science and modern technology are coming together to reshape the future of insurance. John Johanssen, a senior technologist with decades of experience, shares why actuaries are uniquely positioned to lead this change and why they have become some of his favourite partners in innovation. 

From real-world examples of more granular catastrophe underwriting to lessons learned from projects that took longer than planned, the conversation offers a candid look at what works and what does not. We dive into how AI and modern development tools are dramatically speeding up software delivery, where the hype around AI falls short, and why narrowly focused use cases deliver the most value. The episode also tackles governance, security, and what business leaders, technologists, and actuaries can do today to prepare for a rapidly changing, platform-driven future.

Mixed & Edited by Next Day Podcast

info@nextdaypodcast.com

Transcript

Hi, this is Steve Abel. I'm a global technology partner at Oliver Wyman Actuarial. And we're here to talk to various leaders about the intersection of actuarial science and technology. And I'm joined by John Johansen, one of our senior technologists with a hot take on what we're seeing in platforms, software and technology. 

John, do you want to give us a little bit of background on who is John Johansen and what brings you here?

John Johansen 

I am very happy to do that, Steve. Thank you for the opportunity to talk today. 

As you may know, I come to this conversation with about 30 years of experience in helping insurance, insurance organizations think through how they want to prioritize their technology investments. I've done that with a variety of organizations over that 30 years. But really what I enjoy and what brings me to the conversation is, summed up by actually a cover from Wired Magazine of all places where I really feel like I was born at the right time.

The mysteries of software caught my eye when I was a little boy. I still love it, the mess of it, the code, the toolkits, all down to the pixels and the processors. I started writing business applications when I was 13 and I haven't stopped. So I started a software company in college writing code for university departments. I did all those things consultants are supposed to do.

Helped clients succeed, went to work for Big Four, grew a software SI firm from nothing to over about 125 people. Sold that, went into strategy and &A, and now I found a home at OW because they let me do what that 13-year-old did. I get to imagine how modern technology, skillfully applied, can improve businesses, including our own. I get to think about that just about all day, and I get to lead talented teams that make those improvements happen.

Steven Abel 

So give me the connection, why actuaries? Like you could work anywhere. You could work at an engineering firm. You could work at a different kind of consulting firm. What landed you in an actuarial consulting organization?

John Johansen 

I'm going to go back in time a little bit and tell a little story. I was speaking at IASA's CIO forum and they asked us to do a TED Talk. And the subject of my TED Talk was an underwriting class of one. And basically thinking about how technology was helping us move to finer grained underwriting across the insurance world.

In that talk, said facetiously in the voice of a member of the audience who I was pretending was arguing with me, but what about the actuaries? What are they going to say about your premise, John? And my answer was then as it is now, the actuaries, the actuaries are my people. Almost all of our great projects combined forward thinking actuaries, innovative ideas and great technologists. And that combination led to some fun projects and to some groundbreaking outcomes.

It's really, when I work with actuarial teams, I find that we're working with some of the smartest folks that think deeply about how insurance ought to work. And they're fairly quickstudies when it comes to the kinds of things, the kinds of impacts that technology can have.

Steven Abel 

Before we get into your take on platforms and modern tooling, you mentioned some projects. Do you have a couple of examples you might be able to share? Something that went really well and maybe something that went horribly wrong?

John Johansen 

Horribly wrong. So I mean, through the years, there's been a number of examples where we've done a really interesting job applying technology.

The one of the projects in particular had to do with really increasing the granularity of catastrophe underwriting. was one of those that was sponsored by an actuarial team. And what we were able to do was shift a carrier from going from underwriting portfolios of catastrophe risk to underwriting individual catastrophe risks. And we used technology to do that to kind of increase the granularity of those analytics. 

And it really helped them rebalance their portfolio and make their application of capital quite efficient. You know there's sometimes when you say something that's gone horribly wrong sometimes things just cost a little bit more so maybe that's going to be my example for something that went horribly wrong. We got on a thread with a global insurer and we were helping them really think expansively about how they needed to process their business globally.

And we started a project that we thought would be about you know, six months of work and well, we had some really good ideas at the end of six months and we said, well, maybe this could be nine months of work and we had some more ideas and the management team kind of said, okay, okay, okay. By the end of about a year, they were like, all right, let's stop. Let's stop spending on this. They got the outcome they wanted and they had some great ideas along the way, but it was a little bit longer than maybe they signed up for at the outset and maybe it cost them a couple extra dollars.

Steven Abel 

So speaking of taking a long time to do things, is the time horizon to developing sort of modern solutions, has that changed at all in your experience, let's say in the last two or three years?

John Johansen 

I think that there's an emergence of a way of working in terms of building software that challenges a lot of assumptions that building software has to take a long time.

So the answer to your question is yes. We've seen things with modern tools. We've seen modern approaches brought to those projects and they happen just fundamentally faster. We are seeing that, especially in the last two or three years, the idea that we're going to start a software development project by writing everything down fall by the wayside.

And what we want to do is we want to always want to have the right level of documentation. Always want to make sure we've got the right level of governance around what it is we're there to do. But let's get right to building and the new tools go really streamline the process of going from a wireframe design to a working prototype to a version zero to an MVP and then to an industrialized application. And we're seeing every step in that process get shorter and shorter and shorter given some of the tools that we have available to us today.

Steven Abel 

The other day you were sharing an example that might be interesting about an engineering firm that we work with trying to develop some working tech and then you using AI on a similar problem. Can you share that example? I thought it was particularly interesting.

John Johansen 

Yes, I'm happy to. It's another good example on where some of the new tools, AI in particular, can bring orders of magnitude, speed to development processes. We were doing some work just getting to know some software that one of our engineering partners had built. And our habit is to use AI engines to tell us what the software is doing, give us a good summary. It's just a good way for us to get some documentation on the process.

In the code, the engineering partner had said, when this becomes robust, we're going to need to recode this this way. And basically, they had a bunch of to-do items in their software code. The AI read the code and it said, here's what this does. I'm going to write the code that was implied in all these to-dos. 

Do you want me to embed it in the project right now? And it did. So in about 15 minutes. We had all of our to-do lists from our engineering partner completed in the application. It was pretty wild.

Steven Abel 

That's super cool. So you've had a of successes working with AI. Where have you personally struggled? where's the hype? Because we hear a lot about things that work particularly well, but everything can't work well all the time.

John Johansen 

I think a lot of the hype and one of the things that I think we've learned is if you try and solve too big a problem, if you open the aperture on the problem too wide, the AI can't really get a handhold and really help. It's just too general.

And where we see people applying AI effectively, they're doing things that are very narrowly focused. They're really closing that aperture and say, I want help with this specific problem and I want you to perform this specific role for me or do this specific analysis or summarize this specific code. 

And that is where we're seeing people have some terrific success where we see AI struggling is people asking too general a question. I want to ingest any document that might come into my business and I want you to read it and tell me what it is and act on it. It's a very general take and the AI is going to have trouble doing that with any degree of reliability that's going to have it be better than having a human do that.

But where we see success is by aiming at very small problems and saying, I need you to look at the way this calculation is done and make it more efficient for us. I need you to look at this kind of data set and know that here's the things that you're going to find. Look for these and find them in this data. Those are the kinds of problems that we're having great success with AI solving for us.

Steven Abel 

So do you have any examples working with actuaries of problems that we're working on right now or have recently worked on that apply that method that you've just talked about?

John Johansen 

Yes, and one of the things that we have found incredibly effective is to take some of these actuaries that I referenced before are fast studies when it comes to new technology kinds of techniques and give them access to the AIs themselves and see what they want to build.

And one of the things that our actuarial teams have decided they want to build is something that streamlines the process of taking raw data files, data files that have maybe different formats from a single client and conforming them into an analysis.

And we've got actuaries that have built AI applications or used AI to build applications that contain AI to take that process and streamline it, to go from loss run files and really take it through the process to get to where they start doing their value added actuarial things, but really eliminating the actuarial forever, talking about the 80 % of the data engineering they need to do to get to the 20 % of analysis time. And what we're really working on and what they're working on is streamlining that process to go to take that 80 % and make that 80%, you know, 90 % smaller.

And that's the exact kind of application that we see teams working on that I'm going to hang up and literally talk to folks about.

Steven Abel 

As a technologist, do you see the actuaries that surround you changing the way that they're looking and dealing with technology? And tell me more about, if you are seeing changes, what are the changes and why do you think they're happening?

John Johansen 

I think that there's the actuaries have been on the receiving end of a lot of technology advice and seeing a lot of the limiters of technology that have kind of been in place over the last 20 years. Actuaries have been big sponsors of something I call an analytics project that turns into a data movement project. And they've spent a lot of money.

just on data movement and moving information around their organizations. And I think what the actuaries are seeing now is that a lot of those data movement problems are getting solved by things like cloud databases, by some of the investments that organizations have made in data lakes. And they're starting to see that every data project does not become, does not have to be a data movement project. 

And in that way they're just starting to open their eyes. They're starting to be a little more open to saying, wow, this actually might help me do what I need to do. Because I think they've been, if not burned, certainly disappointed by folks that have said, wow, new tech is gonna be a game changer for you. But it never really landed.

And I think right now we're at a bit of a unique moment in time where it's all kind of landing for the kinds of solutions that actuaries need in order to improve the way that they serve as clients in order to prove the analysis that they're doing for clients.

Steven Abel 

So tell me more about this unique moment in time and solutions that actuaries might be involved in or need and work with. What's changed? What's changing?

John Johansen 

We've covered a lot of maybe how people see what's changing, but I've actually, I've said this for probably two decades now that we've been at a unique moment in time. Probably 15 years ago, what I was thinking about was that, you know, we had invested a lot in data and data solutions and the data that we needed, we now had, because if we think back well in the past,

You couldn't really use hardly any institutional data for analytics. didn't have high enough quality. There's been huge changes in tooling in a lot of dimensions. The first is fairly sophisticated machine learning tooling that used to be academic, that used to be university, that used to be incredibly expensive hardware to run on. Those things all came down market. All of a sudden, they were bundled with Microsoft's software.

Each of these things is accelerating in a way that we certainly have access to more data than we can make sense of now. So we don't want for data. But all the tools that we now have access to make that data available to us more easily. 

And the power of the tooling that forward thinking organizations are making available to their teams actually changes the conversation from, we're giving our actuaries end user compute tools to we're giving our actuaries end user platform tools, meaning they're, getting like end user platform development tools because, we're seeing organizations make available databases, make available compute. 

I mentioned earlier that we're pairing our actuaries up with AI engines that help them build software that's going to help them build repeatable solutions. And if you think about that, this moment in time now says, I've got a forward thinking actuary that isn't afraid to build some software. 

And I've got a forward thinking IT organization that marries that up with a scalable platformthat can be available to the entire organization. And at no point do I have to go out and procure hardware. At no point do I have to talk to people about what software am I running on what version of what server. And a lot of those walls have fallen down. And it really goes to, you alluded to a bit of a hot take that maybe I have on platforms.

I'm a believer and there's more than me running around the market saying this, that it is certainly in sight that a lot of the, what I'll call large scale enterprise software vendors really ought to be thinking about what their business model looks like in five years. 

Because if we can all create on things that are end user platforms, then we don't need to pay millions of dollars in license fees every year to enterprise software providers because we can create, launch, harden, industrialize applications that do just what we want them to do, no more, no less. And then we can control that path and we can take that seven figure licensing fee. 

And, you know, maybe we spend a little bit less than that in a particular year but we're gonnaspend that on exactly the kinds of improvements that we wanna make to that platform. And I can absolutely see a day, it goes back to a phrase, everyone's kind of overestimates the change they'll see in a year and underestimates the change they'll see in five years. I think this is one of those areas that we're probably underestimating the change we'll see in five years.

Because I do think that enterprise software is ripe for disruption for exactly some of reasons we've talked about.

Steven Abel

I mean, if you could build whole applications or platforms, I mean, if I'm an actuary, you're in that, that sounds exciting. If I'm a technologist hearing that, that sounds terrifying. Because the old problem was how do I get rid of all these Excel and user computing things to the technologists that might be listening or CIOs or CTOs? Should they be afraid? How should they deal with this sort of explosion of what you're describing, maybe like nuclear end user computing.

John Johansen 

Well, I actually think that a lot of the forward thinking CIOs, lot of forward thinking CTOs have already turned the corner on this and they know that their role is not one of kind of lockdown, you know, provide tight, tight boundaries. Their role is one of enablement. Their role is one of, okay, how do I bring these kinds of capabilities to the teams that want them and can leverage them?

And also, how do I bring what my team's been great at, which is taking those early developed pieces of software and making them bulletproof and making them so that they run in a corner unattended. Because that's something that actuarial teams, that business teams are not going to be very good at.

The good news is that if you think about that as an ecosystem, I've got forward thinking technologists thinking about how am I enabling the platforms that are going to be used by my end user platform developers. And at what point do we turn that over to our teams and then allow our technology teams to operationalize and industrialize those platforms.

That's exactly almost exactly having everyone do what they're good at. Everyone do what they're great at. And it and it really does create this virtuous cycle where the actuaries in our case are getting exactly the tools that they want. But they're getting them delivered ultimately in a way that has them industrialized, that has them scalable, that that allows them to really become assets of the organization.

Steven Abel 

So what are the implications to like governance? If you're in IT or you're in security, and what are the implications to governance if you're just a practitioner today, a business practitioner? What changes do you see coming down the line?

John Johansen 

I think that, I think that what's happening in, in governance in particular is you're starting to see, as you get more people at the table, more people are going to have governance conversations. More people need to be educated as to what does an appropriate solution looklike? What are the guardrails? What are the bright lines? And how do we, as an organizationcommunicate what those bright lines are? 

And also, you know, there's a process that I call running with scissors. We want to make sure that we don't let people run with scissors. Maybe we don't let everyone run with scissors, but maybe some people have proven that they can be the exception. And we do, you you do make thoughtful exceptions on some of those things.

But we see the governance processes as being getting democratized in just the same way as that code development is, as that solution development is.

Data security, I think it's one of the areas where tooling is improving every day. The code scanning kinds of things that go on within our software infrastructure are exactly the kinds of things that are going to make sure that these AI engines are not introducing backdoors. These AI engines are not doing things that are somewhat nefarious, that there's not some kind of leakage of secure information through any application.

So I think that there's tools to the rescue, but I think it's another area where just publicizing to people and saying, here's the rules of the road. Here's the things that you can and can't do and allowing them the freedom to do the things that are allowed is an important part of taking advantage of all the tools.

Steven Abel 

So can you compare and contrast organizations that have a governance method where it's just, no, don't use these tools with organizations that are maybe more nimble and guideline focused as you've described?

John Johansen 

Well, we do work with clients. And I was recently visiting with a client that had a very, very limited view of what they would let their people use AI tools for and what they would let their people use even some of the cloud data tools for.

You know, we sat with them and we said, OK, well, you're really struggling with some problems that create issues for your business. And if you had a little bit more of a liberal view on how we might use some cloud data management tools, you might get a faster answer on some of the things that are expensive for you today and sometimes look pretty insolvable.

So we're seeing that it does cost money to be restrictive. And people that are taking a little bit more experimental view to some of these tools, people that are listening to their practitioners and saying, when they're saying, here's the tools that I'm using every day, here's the problems I'm facing, and here's a way that I've used this tool securely as their way for us to expand its use in some useful way. Those are the folks that are getting area under the curve, they're getting lift from the tools today. 

And just as having a very strict lockdown environment makes some solutions a lot more expensive, we're starting to see the other side of that coin and folks that have more permissive environments are getting those efficiencies much more quickly.

Steven Abel 

I've talked to clients that have told me that they cheat. Like their IT says they can't use AI, so they're using AI on their personal devices. Is that a thing? What are you seeing in the ecosystem?

John Johansen 

is absolutely a thing. It's one of those things where, if, if you're sitting in a technology department or as a chief security officer and you're saying, no, there's rules against that in my organization. I'm sure our people don't do that. Well, I have news for you. They, yeah, your, teams are doing that. 

And, it's really one of the things that, you almost need to make sure you're taking measured risk on because you need to calculate the risk of locking down your environment, knowing that it's going to flow to other places versus providing tools and techniques that your teams can use. Albeit maybe it adds a little bit of risk to your internal tool set, but it's certainly less risk than having your teams trying to go off site and roll their own.

Steven Abel 

So you painted a pretty cool picture in five years of a completely different ecosystem. But today, it's a two-parter question. If I'm a business practitioner, what should I do to be ready for that future? And then the second part is, if I'm an IT leader, what do I need to do to readymy organization for that future?

John Johansen 

So the business leaders need to be thinking like business people, like they usually do, about what specifically tools that they pay a lot of money for are adding to their process. I was on the phone with someone just yesterday and he said, I'm paying X.

And I feel like I've got a very small thing that I need done and it is not aligned. You know, what they are really doing for me as a SaaS vendor is not aligned to what I pay them. And I think that that's a business executive thinking about it exactly the right way. What is the economic value of what they're providing me? And how do I decompose that into other pieces that maybe we can build on our own. 

What is the, what's the real kernel that is uniquely being provided by that particular softwarevendor? And if the business people are thinking about that and they dust that conversation off every six months, every year at planning time, I think they're going to be well positioned when in three years, someone says, by the way, we're now at a point where we think we can run a discounted cashflow on our own or run roll forward calculations all on our own and then they'll be ready to strike.

The IT organizations need to be doing the kinds of things that they've already been doing, which is focusing on that enablement word, focusing on making sure that their teams are always going to get the tools that help them advance. And at the same time, thinking creatively about some of the things, Steve, that you brought up around governance, security controls, and making sure that they're not, you know, we're not going to let the horse leave the barn, just because we're trying to get the most out of some new tag.

And I think we're seeing that. there's organizations thinking about how do we provide cloud platforms as a service? How do I provide databases as a service and processing platforms as a service? And how do I put portals in front of that so that I can request those things and it kind of spins up for me in an unattended way. And now as a practitioner, I've got a sandbox that I can work within. That's way more capable than the SharePoint share that I've been using to manage my business for the last 12 years.

Steven Abel 

What are you doing in Oliver Wyman Actuarial to prepare us for that future?

John Johansen 

Well, I participated in a number of teams that are focused in unsurprisingly on a lot of the things that we've covered today. There's the AI task force that I think you referenced earlier in this discussion is really it's made up of a cross-functional group of people across Oliver Wyman Actuarial that are thinking about how can AI impact what they're doing and not only thinking about it, 

But to the piece of the conversation we had earlier, actually building working software to demonstrate to folks, here's what we think we can do with these tools. So I worked very closely with that team. I mentioned I've been building software for a few minutes and that kind of focus on product management is something that I bring to that conversation and it's exciting to see what the other folks on that team are able to imagine and are able to build.

We also have a tremendous data literacy program that is increasing the skill set of every single one of our practitioners. And that is an investment in making them the next generation of actuaries that are gonna be tech enabled, that are gonna be very, very comfortable playing in the spaces that we've talked about in this conversation and exploiting them and taking them to the next level. So we spend a lot of time in data literacy. We're also building within our organization what we're calling the next generation of product management skill.

It's one of the things that we're seeing is now a core skill requirement for some of our practitioners, much in the way that at some point in the past, everyone needed to be a good project manager. At some point in the past, everyone needed good data skills. We're really starting to see that people that can develop a vision for how an application, how a function, how a business process can be automated and can be performed and then make that real communicate with the builders. That's a real that's a real benefit. And I think in those three areas, we're spending a fair bit of time.

Steven Abel 

John, we've covered a lot, covered a vision of the future, what we're doing at Oliver Lyman Actuarial. If our listeners have made it to the end of this podcast, what are three pieces of advice you might give them?

John Johansen 

Well, the first is maybe the most important and is be curious. One of the things that has really served me well and served my teams well is always thinking about, well, what if, and kind of not being afraid to try new things. And that curiosity can be filled in a lot of ways right now. 

And it's super interesting to try and look at new tech and how it's being used. So that curiosity is always gonna serve you well. I would also always be thinking about information security because your organization's information is a competitive advantage.

And there's a lot of ways that one of the things they say is space is always trying to kill you. Well, all that information is always trying to be a security breach. So always bring that lens to it. And your executives will thank you. 

And the last is have some fun. I think one of the things that. If you listen a little bit to the story I told at the beginning, I get to go to work every day and do something I had a ton of fun with doing when I was a kid. Well, that really makes for short days, makes for short years, and makes for some exciting projects working with interesting people.

Steven Abel 

John, thank you. This has been a very interesting conversation. If someone wanted to reach out and continue that conversation with you, how would they do that?

John Johansen 

Easiest way to get me is through email. That's my first name dot my last name at oliverwyman.com. So it's j-o-h-n dot j-o-h-a-n-s-e-n at oliverwyman.com. I look forward to hearing from folks.

Steven Abel 

Thanks, John.