Scouting for Growth
00:00:16 · Sabine VanderLinden
[00:00:16] S0: If your underwriting team spends more time wrestling PDFs than pricing risk, you are not alone. On a typical day, a single 50 megabytes email can land in an underwriter's inbox bursting with spreadsheets, scan slips, and missing fields before anyone gets near a radar. And yes, much of it still gets rekeyed by hand. But here is the upside. Carriers that get AI right are already pulling ahead over the last five years. AI leaders in insurance created 6.1 times the total shareholder return of Lagardes. That kind of gap changes careers and combined ratios. Today on the Scouting for Growth podcast, we are going to delve into the world of agentic underwriting. The move from manual triage to autonomous multi agent systems that orchestrate intake, risk profiling, pricing, and compliance with human oversight. Think AI colleagues that pause submissions, surface the right third party data and route complex calls to underwriters while continuously learning from outcomes. Our guest is Amrit Santhirasenan, co founder and CEO of Hyperexponential, also known under the name of Hx, an actuary and software engineer who has built the AI native pricing and underwriting platform used by leading specialty carriers. Amrit joined me last year, and I am so excited to welcome back Amrit today for an update just before ITC Vegas, the leading global insurance conference in The United States. Hyperexponential reports 50% faster submission to quote, 10 time faster rate of build and deployment, and supports 45,000,000,000 of GWP annually across 40 plus enterprise customers. These customers include names like Aegis, London, Aviva, Beazley, Conduit Re, Convex, HDI, Markel, and Long Tellery among the so many others. Why should an executive lean in now? Yes, that is an important question. Because the baseline is broken. 86 of underwriters still spend two or more hours per day on annual data entry. 95% of US insurers say their pricing technology needs improvement, and over a third report losing business due to slow or inconsistent pricing. These data set come from research that Hxponential put together for The UK and The United States. That is margin left on the table and broker relationships strained before the first endorsement is signed.
00:03:37 · Sabine VanderLinden, Sabine VanderLinden, Amrit Santhirasenan
For you, trans fascination leaders, under board pressure to prove ROI, scale beyond pilots, and de risk innovation, This conversation is built just for you. In the next sixty minutes, Amrit and I will cover, one, from attachment soup to decision intelligence, how to turn messy submissions into structured signals your pricing model can trust without claring an army. Two, what agentic really means. Multi agent architectures, the agentic AI mesh, and the human in the loop controls executives need to have for auditability and speed. Three, the specialty playbook or the PNC insurance playbook, actually, where agentic underwriting is ready today and where it is notched, plus the metrics. Executives should track second time, hit ratio and loss ratio uplift. If your strategy is to move from decks to deals, from rekeying to real time, this episode is really for you. It will show you what good looks like and how to get there faster. So let's dive in and welcome Amrit to the Scouting for Growth podcast. [00:05:14] S1: Hello, hello, hello. Amrit, I'm so pleased to have you back with me. [00:05:21] S2: Thanks for having me, Sabine. It's good to be back. [00:05:24] S1: So let's just go straight into our conversation because I want you to start our conversation with describing one of your experiences actually at loads of London. I know you come from loads. And maybe perhaps going and diving into one of those 50 megabytes submission emails we used to receive at the time. What has changed?
00:05:48 · Amrit Santhirasenan
[00:05:48] S2: Yeah. Great question. I always love reminiscing with... About Lloyd's with you. So I come from the era where I would see the brokers and claims handlers turn up in the lift with files in a suitcase. So the email submissions, that was a luxury at the start of my career. So it's amazing to think about how much it's changed today, hasn't it? You know, particularly with the kind of dawn of AI or generative AI in terms of untangling the vipers nests of these submissions. Right? So we've gone from an era where data was, you know, bound literally in physical PDFs all the way through to email based submissions, people doing work much faster, still very manually through to the breakthrough we've had with LLM's ability to synthesize this data at pace. And the thing that's been most exciting for me, I'm a self professed, very proud nerd, is just to look at the pace at which the kind of capabilities of the core models have got so good that, you know, in the space of six months, you know, what historically, you know, even six months ago was a whole product and capability and feature set is now, you know, within the gift of Gemini or, you know, GPT-5. Alright. Or... Oh, lord. It's just... It's amazing really to see how much the change has come through.
00:07:01 · Sabine VanderLinden
[00:07:01] S1: How much change has come through? And I use all of them because I want to compare the result. And so it's fascinating, the speed of change, and I think it's important for us to realize that. So let's dive into specialty underwriting, right? We are in the age, as you said, of ChatGPT, Claude, Gemini, amongst others. Tell me, what is a genetic underwriting for you? How do we move from the chaos of spreadsheets, PDFs, and emails within the under... Underwriting sphere? And why do you believe artificial intelligence is now crucial to manage data complexity in our industry?
00:07:43 · Amrit Santhirasenan
[00:07:43] S2: Yeah. Yeah. It's a it's a great question. And, you know, you'll know that, you know, although we sell and service more than just the specialty markets now, our home, our core is the specialty space, and it's something that's personally very close to my heart. So I often reflect, right, if you were to think about your dream technology for a specialty in commercial underwriter, right, you've got this heterogeneous data. It's disgusting. It comes from a 100 different datasets. And actually, the data that an underwriter and an actuary can come together is often the source of differentiation. Right? And I think if you were to think about the dream technologies to untangle all of that, it would be something not far off what we see today in these LLMs. If you're a high throughput, high velocity SME or underwriter, the state of the art was pretty impressive. You look at some of our clients, likes of Aoife, they're pressing risks extremely quickly, even before Gen AI came along. The breakthrough technology, I think, for specialty and commercial insurance to really break the back of that extremely of extreme morass of data is the Gen AI, right? The ability to structure it, to synthesize and analyze unstructured data from a variety of sources in a conversational way, in an exploratory way. Right? You know, that's the thing that I see really clearly. You know, to build on that question about why is this important? Well, think, you know, and what's agentic underwriting? Well, you know, the thing I tell my team is in the same way that maybe fifteen years ago, no one would have imagined us doing all of our banking, living our day to day lives on one of these. I think there are lots of people that are really underestimating how much they will do their day to day jobs in symbiosis with a set of agents. And so for Hx, one of the ways that visions that we have is for us to be the leading frontier provider of underwriting and actuarial support agents. Because that's just where the market is going to go. If you're an underwriter filling out a spreadsheet or a renew model, right, you know, a spreadsheet maybe 1996, maybe 2023 as a Renew model. Twenty twenty five is working with the Hx underwriting or actuarial agent inside a Renew model to have your peer partner to help you get to the best result.
00:09:55 · Sabine VanderLinden
[00:09:55] S1: Can you tell us, you use a really key term which I was using just last week with the team at Microsoft. You use the term frontier. What does exiting frontier means for us in insurance?