Ash Sawhney On The Future Of Underwriting And AI

Paul Tyler (00:00) Welcome to the LNA Lab, where we take a close look at what's actually transforming life insurance and annuities. I'm Paul Tyler. Each episode, we sit down with operators and innovators to unpack their lab notes, the bets they made, the problems they solved, the results they saw, and what they would do differently next time.

Hi, this is Paul Tyler, and welcome to our first episode of a new podcast. And thanks for joining us. If you may have followed us from other shows, want to welcome you and hope you certainly like this, follow us and join us. And today I have a great guest, Mr. Ash Sawhney. Ash, welcome.

Ash Sawhney (00:42) Thanks Paul. Thanks for having me.

Paul Tyler (00:44) Yeah, it's funny. I go to every conference and people, I know Ash. Ash, you have established a broad following, I think, among the industry. I certainly, your reputation proceeds, for me, proceeds our working relationship. But maybe tell people a little bit about, you know, who are you and what do you do here at Zinnia?

Ash Sawhney (01:06) Yeah, thanks, Paul. So my title is President and Chief Revenue Officer. Really, what I do is help establish relationships with our clients and foster those relationships.

To put things in context, Zinnia is an insurtech firm. We are firmly entrenched in the life and annuity space. What we do is digitize and automate literally every aspect of life and annuity processing right on the front end from agency management all the way through how we match clients with the right products or needs analysis, quoting, illustrations, order entry. We have different flavors of electronic applications. We underwrite policies, we issue them and we administer them.

So it's truly an end-to-end value proposition and we are doing this at scale. So we have thousands of advisors that use our platforms. We run literally millions of illustrations a year. We have more than a 50% market share in our order entry space of the annuity side. We have the most modern policy admin platform that we've launched. So we try to pick our spots, but we also try to be the leaders in that space.

So that's a little bit of what I do and the firm that I am with now. Prior to that, Zinnia acquired eBix and I was running the life and annuity business at eBix before joining Zinnia.

Paul Tyler (02:40) Yeah. Well, I guess in terms of customers, if I'm an entrepreneur who has raised $20-30 million, want to start an insurance company, should I come knocking on Zinnia's door?

Ash Sawhney (02:52) You do because we provide everything that you need to process business, right from helping you set up your products in an admin platform to help connect you with distribution, enable you to submit applications, enable you to underwrite policies. So we offer all aspects of what you would need as a new co, for sure.

Paul Tyler (03:14) Yeah. Yeah. Well, why don't we just talk a little bit about like that. I mean, you know, Zinnia solves problems with a lot of big companies. But you understand the challenges of actually being a small company, starting a small company and then serving big companies. Maybe just talk to us about how you, what was your journey? How did you actually dare to start your own business?

Ash Sawhney (03:38) Yes, that's a good question. Yeah, I started something back in the early 2000s and I actually worked in the industry for a while and got to learn a lot about systems projects and a lot of projects were failing at that time. And I saw a lot of challenges in sort of the new business and underwriting space.

It never used to be a discrete category. New business underwriting was always embellished within policy admin platforms. So we saw an opportunity to carve that out into its own category. So we formed a company back in 2000. We scaled it for many years. Sold it to eBix in 2012. So yeah, that was my stint as an entrepreneur.

Paul Tyler (04:22) I guess, let me ask you just a little bit of that. What was the scariest moment as an entrepreneur and how did you get through it?

Ash Sawhney (04:30) You know, it's interesting, Paul. It's fun. It's a lot of excitement, a lot of exhilaration. It's also very frustrating. You know, being in the insurance industry, everything moves very slow. So you come out of a meeting so excited that you might have won a new client and you know, you're still hoping after six months, things move very slowly. So it's like I said, a lot of blood, sweat and tears, right? And a lot of good luck is needed to succeed.

Paul Tyler (05:00) Well, it was interesting. We were talking to an analyst, well-known analyst in the business. I won't actually mention his name because I think he may have gone to one of our competitors recently. And it was interesting to ask, you know, he's been working some cutting edge technologies. I remember PlanetSoft. That was leading edge, you know, when I started looking at it working with it.

I guess if we fast forwarded, you know, closer to today, how's underwriting evolved? You said it was really kind of embedded in the policy admin systems. It sounds like you invented a category really with your product. What's happened since then?

Ash Sawhney (05:42) Yeah, so we were one of the early stage firms that sort of were in that category, right? And some of them didn't survive. We were lucky to survive.

But yes, so underwriting has evolved quite a bit, you know, I would say in the last two decades. And it seems to me like it evolves every like five years. And I don't know why it's five years, but that's how it seemed to be.

In the very early days, you know, the trend was towards digitization, right? So previously, underwriters would do everything manually, spreadsheets, had manuals sitting on their desks. They would have note sheets where they would literally do debit credits on risk assessment. So a lot of that started to get digitized into rules. That was the first evolution.

Then paper started to give way to data. Electronic applications started to become more prevalent, which ultimately helped in the underwriting process, especially bringing more underwriting to the point of sale.

We also saw a lot of advancements because of ACORD. These systems don't exist in a vacuum, right? So new business underwriting gets tethered to practically every other system within an insurance company. So you have to have a very robust platform and a strategy of how to integrate into the ecosystem. We developed what we called an integration hub that allowed us to normalize and integrate into a myriad of different platforms, policy admin systems, imaging systems, workflow systems. So that was a big movement.

The reinsurers had a big part to play. They started to take their underwriting guidelines and make those available in systems and platforms. So we started to integrate with those expert systems.

And then finally, data started to become more accessible, more real time. It used to be that requirements, medical evidence requirements were all like manually ordered and manually received. Sometimes they would come back as paper. So the industry saw a lot of that get automated, real time requests, data coming in electronically, sometimes real time, and then the analysis of that got automated.

So those were some of the advancements, I would say, that we saw during the last 20 years.

Paul Tyler (08:10) Yeah, and I guess there are certain trends that seemed to, I thought, would take off and they sputtered, at least in the last couple of years. Like one was not electronic medical records, I guess electronic medical records. Actually, I saw a lot of vendors selling complete almost like, and I'm not going to use the right term for it, like Paul Tyler's medical records from this particular region.

It felt like there was a lot of promise, but every time I talked to a vendor, well, they had 80 percent hit in this area of the state, but they only had X percent on that. Maybe, I don't know, can you align here? Why has that struggled and do you think that's gonna change in the future?

Ash Sawhney (08:52) Oh, it's changed a lot already, Paul. Like, you know, I think the best example I can give is prescription data. The hit rates used to be really low. And now prescription data hit rates are really high. So electronic medical records are evolving, but there's a lot more richer data sets that are available now than they were even like three or four or five years ago. So big advancements in that space for sure.

Paul Tyler (09:18) Now, the elephant in the room is AI. Yeah. And underwriting is one of those areas where, wow, the regulators are deadly focused on it, but the industry is also equally focused to say, what are the benefits? How do you see AI permeating the industry today? Where do you think it'll be in the next, you know, what do you say going on the next two to three years?

Ash Sawhney (09:44) Yeah, listen, AI is obviously impacting every industry and insurance is no different. And it's going to have a major impact. It's already starting to have an impact even in the underwriting function.

But if we think of this more broadly, Paul, before we talk about insurance and underwriting, I think we need to see how AI is generally impacting healthcare and medicine, right? Because at the end of the day, some of those research and those initiatives have an impact on life expectancy, which ultimately has an impact on how we assess risk, right?

So lots of really exciting stuff going on there, which is way past what I would say, just like lab tests. They're now more real time in use. Drugs are getting launched faster. AI is having a major impact in how drug trials are conducted. What used to take years is now taking months. The success rate of those phase one trials has almost doubled with AI because you're able to match the right trial candidates and find the right profile.

So all of these drugs that are coming out that help with life expectancy, those are going to have a big impact. You look at AI and imaging. This is a big area where we've seen improvement. AI can detect patterns in images much better than the human eye. And that has a profound effect on detecting some serious illnesses early on.

Genetics, another big area of study where AI is using large data sets for DNA sequencing and helping diagnose and cure critical illnesses. So a lot of these things have downstream impact on our industry.

You look at genomic studies, which is helping us launch very targeted medicine. So it's no longer like one size fits all type medical cure. It's very targeted medicines.

We also see, as an example, home health or remote monitored healthcare through AI agents. That means that success rates are higher, people are being monitored and you can early detect serious conditions. We all have seen like wearable devices and how that's impacting healthcare, glucose monitors and heart monitors. So there's a lot of data, biosensors, right? A lot of data that's coming in that's gonna help with assessing a person's risk.

Full body scans, that's another example, much clearer outputs that come from those AI enabled full body scans. Lots of advancements also in pharmacogenomics, right? So the efficacy of drugs has improved so much because you're able to align those drugs with the right patients.

Doctors, and here's a really good example, doctors have started using AI scribes. So they're sitting across with a patient and this AI agent or scribe is taking notes. They're sifting out all the small talk, focusing on the real content. So all this replaces what used to be like APSs or doctor's notes, which were hard to circulate around.

So all these things have a downstream impact. And I feel like this is not stuff out in the future. Many of these things are here and now.

What it means from an underwriting perspective, we have to have systems that firstly can take in and ingest all these different data sources. And they can synthesize that data. And that's where AI comes in really handy. You've got disparate data sources, different formats. You have to like holistically look at everything. Much more difficult to do it just manually or even through systems. AI does a really good job of all that.

Now, a lot of these new research and drugs are expensive. So it's up to the actuaries and the underwriters to figure out what protective value they provide and what's the ROI of doing some of those tests. But the fact of the matter is, it's all here and it's available. So those are some really exciting things.

And I mean, it's also, I would say, AI just in the underwriting function is having a very positive impact, right? So firstly, think about generative AI. You have a case that you're underwriting that has so many different artifacts that could have images, could have doctor's reports, they could have a person's profile, LinkedIn profile. AI is able to sort of synthesize that data and make meaningful use in risk assessment.

Agentic AI, being used so broadly across many industries, there's some fantastic use cases in underwriting. This dreaded informal cases that we see all the time in the industry, that can be easily handled by AI agents, right? Many of the manual tasks that the underwriter does can be fully automated.

Think of machine learning and the impact that's having. You can enable underwriters to organize their work more effectively through data models, machine learning to say, okay, which cases are going to be most important today? Is it the cases that are going to be most shopped, the most competitive cases? So a lot of that, which used to be sort of system driven is now driven by machines that are way more savvy.

We look at robotic process automation and anything that's repetitive, anything that's predictable is ideally suited for that. So you think of case management, ordering of requirements, receiving them, creating letters, communicating with the agents, all these reviews, suitability reviews, compliance reviews, all of this gets automated.

There's so many other use cases, Paul, like the one I really like is natural language processing, which can reduce fraud quite significantly. Now you can have, you can look at, for example, if you're taking a Part B, you can have that be done by artificial agents. You can do sentiment analysis. You can analyze acoustic patterns to see if there's any fraud. So these are like some really exciting things that I think are gonna really impact our space.

Paul Tyler (16:10) Yeah, you cover a lot of topics there. It's you have had conversations with people, I'm sure you have to Ash where you say, well, I have to explain life insurance. OK, hold on, sit down, get a drink. We'll stop. I'll stop talking about an hour.

But from an underwriting perspective, I think I heard you say — and tell me if I missed anything from AI. It's a two-part question. So how could AI impact the industry?

One is, okay, clearly the efficiency. I'll kind of go a little bit in reverse. The efficiency of how I underwrite, the more efficient, the quicker it is to do the informals, the quicker it is to do the cases, it means that I can do a lot more with a lot less. Sounds like there are skills, okay, that's going to impact in terms of what that organization would look like.

You said that it would potentially impact the fraud. This was really interesting because people don't understand. There are a lot of fraudulent applications that are put in. It costs everybody a lot of money.

The fourth was, I think you mentioned was just the quality of decisions because sometimes we're so focused on shrinking the time or measurement of underwriting. It's easy to measure. Did it take 25 days or 23 days? But really, what was the quality of the business you put on? So embedded value, that's a tough one.

And the last one, which is really interesting, which is AI just may actually change mortality tables. I mean, did I kind of get the, are those good summaries Ash?

Ash Sawhney (17:50) Yeah, those are good summaries. Like, listen, when there's the real potential of life expectancy changing, people living longer, that ultimately has an impact on how you price that risk. And all of these things that we are seeing in medical science are pointing towards a positive trend in terms of life expectancy.

Data sources are now abundantly available. AI is able to synthesize data. So you're right. It's going to be more accurate decisioning, better pricing, more accurate pricing. It's going to result in us being more efficient when we process business.

Underwriters will always be needed. It's not a profession that's going away. There's sort of the tactical side of underwriting and there's sort of the art of underwriting. And it's the art part of underwriting that will always be there. That's when underwriters look beyond just the data. They use judgment, right?

You can have two individuals with the exact same profiles and medical backgrounds, but the risk profile would be different. And that's where the underwriter's knowledge and what to look for, what trends to look for, is the person's health improving, is it declining, right? What about their lifestyle? What about their habits, right? Those things are taken into account.

Underwriters, they negotiate with advisors, right? So that's a skill set, that's an art. Like when can I be more aggressive and when do I have to give in? Those kind of judgments, complex cases, that's where you're still gonna need the underwriters' help.

But yes, those are things that are changing.

Paul Tyler (19:38) Oh, they are. Well, you know, it's interesting. I talked to Ed Mudd, we were at the AHOU conference. I actually talked to one of our customers. It's interesting. One of the underwriters, she explained to me how different it was to negotiate with an agent in Texas than it was in New York for the very, very different. You know, that's not a skill set you find in a book someplace.

Very last question. For any underwriters or people in the operations area responsible for underwriting, what piece of advice would you give to them as they think about leveraging AI or at least setting their organization up so that they can use AI in the future?

Ash Sawhney (20:20) Paul, I would say treat it as any sort of tech project. You treat a tech project, firstly, you have to establish a good team that should include people with knowledge of AI, people with the business background and knowledge. So underwriters will absolutely play an important part. The technical teams, so you've got to establish the team.

You have to establish a clear goal. What outcome do you want from that exercise? Do you want to reduce the calls coming into an underwriter? Do you want to increase the number of cases that can flow through without involving an underwriter? Can you reduce cycle times by like 50%, right? So the goals have to be realistic, but then that helps you drive towards that objective.

And nothing is better than early wins and success breeds success. So AI is a journey. Find the right use cases that can make a material difference. So that's just one suggestion.

Training data is so critical in sort of the AI world, right? So we sit on a ton of data, the industry does, but it doesn't mean that that data is sort of conducive for training, because when that data was collected, we didn't have AI in mind. So analyzing your data sets, making sure that you have all the variables that are needed to properly train those models, making sure that the data sets are statistically relevant. Also that there's no bias, right? So data samples can very easily lead to bias, minority groups may be disadvantaged or immigrants or certain ethnic groups. So analyzing data and making sure that all the noise and the bias is taken out is super important.

We are a regulated industry, Paul. So it's important that we have transparency. Why did, how did we arrive at those decisions when we get into audit and contestable claims, we have to sort of trace back and see how decisions were made. So that's a very important aspect.

And finally, I would say, this is an evolution. It's a progression and it's constant learning. Models need to be constantly tweaked. But it's super exciting.

Paul Tyler (22:40) Well, listen, Ash, thanks so much for the time. And we'll add the contact information for anybody who wants to reach out to you in the notes of the show.

Thanks for time and look forward to see what the next year brings.

Ash Sawhney (22:54) Absolutely. Thanks for having me, Paul.

Paul Tyler (22:56) Good. Thank you. Thanks.

---

Thanks for listening to the L&A Hub. If you enjoyed the episode, follow the show wherever you listen and leave a quick review to help others find it. Until next time, keep testing, keep learning. Brought to you by Zinnia.