Scouting for Growth
00:05:22 · Sabine VanderLinden
[00:05:22] S1: Wow. Congratulations. You have been in this innovation space leading amazing projects and helping actually, like IAM, move corporations from highlight to execution. Right? Right. So when you look at the larger organization like AXA, for example, you know, 30 old, core system potentially, or maybe more, you know, updated, how do you bring your knowledge from working with big hires into that new world of open AI? You know, how does that look like and how you start solving big problems?
00:05:59 · Bastiaan de Goei
[00:05:59] S2: I love that question. And I think that there are, two specific elements to it. And I think the first one, is rooted in understanding and trust. So one of the first things that you need to nail or really get done is you need to start talking, get into a real conversation. That conversation needs to be built on understanding and trust. So when you are talking about complex large language models, OpenAI's recent Astra models and or like persistent agents, the first thing that needs doing is to translate those incredible capabilities in something that is real for an insurer. And so, okay, there is something that is a persistent agent. What does that mean for an underwriting workflow? And so like truly illustrating that is step one, cause that earns you the conversation and earns you the trust to say, okay, this is a technology company, but they have an understanding of what I experience as an insurance professional. Now the second step is to help guide and work together with large companies like large carriers to actually make that real. And so many times what I find is that I actually have to temper enthusiasm Because as soon as people have an understanding, okay, now I understand that what this could mean for me as an organization, the natural inclination is to then want to do everything all at once. And that is actually the wrong instincts. What I found is that it is much better to basically achieve efficiency by a thousand cuts. And which means start small and add rapidly additional aspects of automation and efficiency to them. And we can work together in a partnership to help make that real.
00:07:48 · Sabine VanderLinden
[00:07:48] S1: So I want to start diving into what you just said. Right? What is already real, right? The intelligence layer. And what we want to do is move that intelligence layer into production. We heard about the numbers, too many pilots, not enough execution. And we are here at ITC Vegas to actually drive, you know, that operational execution. So everyone at this show has seen the demo. I mean, we have a lot of demos, right, downstairs here. And where is AI generally load be real inside carriers today? So that is the first part of my question. Doing the work that would break something if you switch it off tomorrow.
00:08:32 · Bastiaan de Goei
[00:08:32] S2: Right. So I think that's a super fair question, and this is one of the things that I would ask as well when I'm a a buyer. So like where... That sounds great. Where does this already work? Where are you achieving results? And can I talk to those folks? So we are very lucky to work with some of the largest carriers in the world. Some of the public examples that I can share are for Travelers, for example, which is obviously a major insurance carrier and we are lucky enough to have worked with them and still work with them on their claims operations. And so this is a company that obviously has been in the market for a really long time. Very reputable business. It needs to be correct and precise. Right? Insurance, relations, they are all based on trust at the end of a day. And the moment of truth is the claim. That is that insurance is this magical industry where you spend a lot of money for a long period of time and you don't actually know whether it actually lives up to something and this comes real at the time of claim. So what we did together with Travelers was to make the claim process intake, intake process much, much easier. And so today, within OpenAI powered Travelers solution, 80 to 90 of data that as part of the claims intake process is fully done with the AI powered tool done by Travelers. So that's a real tangible example where a lot of these automation and double check and all of that is all done automatically. Another example from a, a different, business line, health insurance, if we take Oscar Health for example, they found that their particular friction point was not on the intake, but it was on the double checking and the actual verification of the claims. And so their analysts and claims adjusters were spending a lot of time looking at all of the different data and verifying that those were correct. And so we developed, co developed a solution with them in order to accelerate that process. And in this case, we were able to accelerate that by 50%. And so that means that every single adjuster just had like 50% more capability, capacity to process claims, which at the end of the day meant that the experience for the end customer was much faster and therefore more satisfactory.
00:10:59 · Sabine VanderLinden
[00:10:59] S1: So looking at both of these examples, you know, what could we say is common when you start looking at your experiences working amongst and with a lot of carriers? So that would be the first part of my question. I suspect you you see some patterns. And and some of the conversation I'm having here at the conferences, how do we move maybe from one siloed workflow to looking at end to end workflows and starting compounding the results we see from our execution?
00:11:29 · Bastiaan de Goei
[00:11:29] S2: Yep. Great question. I think that this actually applies to transformation generally. It is not even specific to insurance. So we've seen actually a lot of companies across, our entire customer base who are actually really good at optimizing a specific step in a workflow that could be data intake, it could be data validation and verification, two examples that we just talked about, could be follow-up, these types of things that are practical, but they're also isolated. And so when we are, as humans, thinking about an end to end process, we are thinking about all of these different steps in one single flow. And that is earlier today, I spoke with a gentleman who said, is this the new straight through processing? And I said, well, you could think about it that way. And why I'm so incredibly excited about the latest technologies and developments is that powerful frontier models like Astra, they are able to stay persistent on their task and see it through end to end. And so rather than taking one system and one process, it is able to understand that for the next step in that process, needs to consult with another system and with another system all the way pushing this towards the end. So if we are talking about a underwriting process, for example, in commercial underwriting, underwriters receive submissions from brokers consisting out of like different data packages, loss runs and broker presentations and things like that. And they obtain these manually, they review them and the end result is that up to, you know, 50% sometimes of these submissions are being ignored because the underwriter is simply inundated by submissions and basically ignores the rest. And so with persistent agents, like Astra, you are able to actually do like a lot of those flows so that it picks up the submissions from the email system. It's able to find and see the data in those. It's able to consult with a pricing engine and to actually get this priced. It can balance this with your portfolio and see if there is... This risk is accretive or not. And subsequently use your company's templates to actually prepare a proposal for this end customer. And so here you are touching like four or five different systems in order to get to the end result of a actual quotation. I think that is what we are talking about when we are thinking about all complete processes.