Watch Christopher M. Wells, Ph. D., Indico VP of Research and Development, and Michelle Gouveia, VP at Sandbox Insurtech Ventures, in episode 27 of Unstructured Unlocked with Abhi Kothari, Practice Director at Everest Group.
Michelle Gouveia: Hey, everybody. Welcome to another episode of Unstructured Unlocked. I’m co-host Michelle Govea,
Christopher Wells: And I’m co-host Chris Wells,
MG: We are excited to be joined today by Abhi Kothari, the practice director at Eris Group. Avi, welcome to the podcast.
Abhi Kothari: Hey, thank you, Michelle, for the warm welcome. And hi, Chris. It’s exciting to be part of this podcast and share my insights with you. I’ll introduce myself if that’s okay. And then we can go ahead. Yeah. Perfect. So, hey, I’m a practice director with the Aest Group. The average group is a strategic research advisory and an analyst firm. And we track the overall market when it comes to the outsourcing industry. We follow both the service providers in the enterprises and also the deal enablers within that. So it’s a complete ecosystem that we deal with. As far as my role is concerned, I deal with the insurance practice, where we cover areas across L N A and the P N C domain.
As far as the my roles and responsibilities are concerned, there, there are primary two work streams that we work across. The first one is syndicated research where we have our own peak rating where we read all the service products in the market for the enterprises. We produce state of the market reports where we give a pulse of the market, what’s going on in the market. We also go ahead and publish lot of thematic viewpoints. So for anything which is going on within the industry. And apart from that, we create service provider profiles also for the enterprises so that they can have a quick summary of all the providers that they wanna engage with. So that’s one part of the work stream that we work with. The second is the customer engagements that we end up doing for the clients, where we help them with various problems, treatments around their G T M strategy, competitive intelligence, account intelligence, or market opportunity assessment, et cetera. So, so it’s a, it’s a multitude of things that we work across, but primarily driven around the outsourcing industry. And that to the area is insurance. P p Ss
CW: Fascinating. Your your viewpoint is quite unique for this podcast. So I’m excited to to dig into some of the reporting that you’ve done recently. But I wonder if you wouldn’t mind spending some time for our audience talking about, you know, what, what’s the process for creating these reports? How do you decide which things are worth reporting on? Who do you talk to? And then how do you, how do you, like, what’s the process of making the sausage to get this thing across the finish line?
AK: Well, definitely, that’s an interesting question that you have asked. So, the process itself of what to cover for the next year starts at the end of the year where we decide that what are the topics that we wanna explore, but our work doesn’t stop there. We continuously go ahead and track the market. What are the recent developments that are happening? So last year when we are preparing this particular agenda recession wasn’t on the cut. People were thinking the decision was gonna come, but it didn’t happen. But then this year we thought that, okay, speaking up a stream, we should do something about it. And we went and created a report around that. There’s a recession viewpoint that we have created for the the industry. That’s just one of the examples that I’m taking, taking apart from that, the generative AI is what is take generating a lot of us right now.
So we are thinking that, okay, we should be writing something about it, right? So what we tend to do is that there are certain areas that we have observed throughout the year, huh, that we see that should be explored for next year, for, to educate the audience, to educate the enterprises and the service providers, so that, that’s one way to look at it. But then we don’t restrict ourselves to just those areas when we see that the latest developments which are happening, and a lot of these macroeconomic factors, which has impacted the industry this year, had been unprecedented. Like any other, if you look at any of the other recent years, there were a lot of events which has happened, right? So we thought that, okay, these are also some of the things that we should be covering when it’s happening, because otherwise you’ll just lose it, right? Last year was really about metaverse. Now no one’s talking about metaverse. Now it’s all about generative ai, right? So it’s all about changing so rapidly, we can’t just wait for a year to come out with that. So that broadly, the process of selecting the topic for any of the <inaudible> that you wanna cover.
MG: And, and I mean once and we’ll get into the details of, of the specific report that we’re, we’re talking that we wanna talk about today. Yeah, definitely. But in, in general, once you’ve identified that theme and, and that space that you wanna pursue there’s obviously some type of competitive landscape analysis that you do, right? To identify the companies that you’ll, you’ll be assessing. How then, once you’ve selected those companies, do you vet them? Do you talk to the companies themselves? Do you talk to their customer base? You know, just, just curious how, how you then take them to, to rank them within the report and, and compare them to one another?
AK: Yeah, so so I think, so there are two kinds of reports that we are talking about. So, so when we talk about peak which is where we rank all the companies and all the service workers, that’s a, that’s a proprietary rating that Everest Group has developed. So there, what we do is, the first step of that is getting the request for information forms out for the service providers. We send them a detailed list of questions that we want them to answer on. And that’s all about their existing business landscape, about their current clients, about their contracts and all, which is there. That is just the first step of it. The next step is that we ask them to schedule a call with us where we will go through all the details that have been shared with with us.
They, they give us a presentation on the future, a looking view of how the industry is going to go about and how they have performed against the industry. That’s the second step. And then we also go ahead and reach out to some of their clients and do a reference survey of how their performance on some of those contracts have been. And that’s also very exhaustive list of questions that we go ahead and ask. We explored all the elements around how the existing contractual performances to whether they would end up selecting some of these service providers in the future. So, so that, that’s an, that’s an overall process. That’s, that’s more external view to it, right? And that’s just the, the initial steps around it. But then we, we feed in all the data that we, we get from the service providers. Apart from that, we also have a market perspective, because we talk to keep on talking to insurance enterprises and all, we use all of that to create a model around it with which we call our peak model.
And this model is, is effectively used to create the peak rankings for the industry. And then at the end of it, we have the ranking where we go ahead and rate all the service providers in three categories, which is leaders, then comes a major container, and then comes the aspirin. So that, that’s about, that’s a bit about the process, how we go about it. It’s a five month long journey. In this process, we end up evaluating more than 20 service providers, at least when I talk about the insurance month. And we make sure that we leave no stones unturned because we want to get, we wanna give accurate picture to the enterprises that this is where the current industry looks at. So that, that, that’s about the peak report that we published. The second thing, which is, which we’re talking about the thematic viewpoints, right?
Some of the, the industry relevant issues that we pick up there, we do, we don’t go as extensive as our peaks. We don’t end up talking with all the service providers, but when we pick beyond where we are picking up any of the topics, we generally have an idea that with service providers are ha are having a good lead in the market. So that’s where we decided that, okay, we wanna speak with couple of service providers on this topic. We wanna understand what they’re doing in the market, that, that’s getting one side of the the industry. But then we also go ahead and speak with some of the enterprise clients. We have a regular connects with them. We speak with them that what are their pain points? What are they seeing in the market? Are they even ready for what is happening in the market? Do they have a different view of, of how the market’s gonna shape up? We combine both of this, and then we have our own average group proprietary research that we keep on doing continuously. We use all of this elements to create a viewpoint, and then we, then we go ahead and present it to the market. So that, that’s about the process when it comes to some of these viewpoints.
CW: Right on. And when you’re talking to folks in the enterprise, what are some of the typical roles or personas that you’re speaking with mm-hmm. <Affirmative>?
AK: Yeah. So so very, very, when it comes to insurance enterprise the lot of times the touch points are initial touch points are the, the procurement managers. We end up dealing the, the procurement heads of the p p s and on. But from there on, depending on the range of topic that we are discussing with, they help us connect with the business stakeholders on their side. So if it’s on the l a side or the p n if the enterprise is dealing with multiple service lines, then we have to end up telling them that, okay, we are looking for certain service lines that we are looking for certain topics. If we can find an expert for that, and they help us arrange for those calls. But our initial touch points are always through those procurement leads because they are the ones who are in, who we are dealing with day in and day out.
CW: Interesting. And beyond the procurement folks, and this is not a lead, I’m trying not to make this a leading question. What’s the insight for those folks to talk to you? Like, why do they pick up the phone when you ask?
AK: Yeah, so interestingly, average group has lot of, in fact, when I talk about the insurance enterprises, it has highest number of enterprises, its customers. So we have two different kind of membership, which is there. So one is outsourcing excellence membership, where they have access to the, the companywide research that we are doing. And that’s the biggest incentive that they have. Even enterprises want to know what other enterprises are doing, what service providers are bringing out in the market. They wanna know that if I’m wanna outsource in particular area, which is the service provider that I should go with versus the other. In fact, a lot of time what has happened is that we get, and they have an option of sending us an inquiry. We day in and day out, get an inquiries from them that, okay, we wanna understand that, okay, we wanna outsource in this particular area.
Do we have a list of service order that we should go ahead and consult with? What other capabilities do they have the, the right set of experience to serve us? So the, the, this is just one set of the question that I’m talking about. They all, sometimes they, when they wanna venture into new area, that time they end up asking us that, okay, we are planning to launch a new product around it. Do you have some research done around it? Do you have a view of some <inaudible>, what they’re doing and all? So these are the typical inquiries we keep on getting from them. So that’s their biggest incentive. So when they talk to us, both of us learn in the process. Yeah. And that’s where this, this is a symbiotic relationship where everyone has to gain from this conversation.
MG: So, so Avi, I think we’ll stop teasing the audience with it. So specifically, the report that we wanted to chat with you about today was the digital underwriting building operational efficiencies across the underwriting lifecycle report, which is a mouthful. But it hits on a topic that we talk a lot about on the podcast, which is automation, workflow improvement, process efficiency as it relates to, to underwriting. So, you know, given that we talk a lot with individuals at specific insurance companies in their process, I’d like to take it a step higher. And just, you know, based on what, what’s in the report, what are the major, what themes or trends that you’re seeing in the market as it relates to these insurance companies and the desire to work with, with vendors for the underwriting workflow?
AK: Yeah, so that, that’s an, that’s a very interesting question. I can speak for it whole day, but I’m sure we don’t have only day to discuss about that <laugh>. So <laugh>, you could
MG: Have said it depends, right, Chris?
AK: I think, I think that’s answer for all the questions I can just get away with <laugh>, I wanna say for those questions, which I dunno answer to, but coming back to, to what you just asked. So, so see, the, the whole idea that we, when we, we went ahead and we wanted to write a report about digital underwriting is because last year before this journey, I came into the picture, one of the themes that we were seeing across the market is that claims and policy servicing had been outsourced, like the digital situation that has came into the, the, the market, right? And what are the new areas that enterprises were thinking about outsourcing. Obviously some amount of outsourcing were happening in those areas, but then underwriting was one of those areas, underwriting and actuarial. But touching upon underwriting, because that’s the topic of this particular discussion, underwriting was one of the areas where we saw that the outsourcing has been consistently increasing in the last two to three years.
So that’s where we, we thought that, okay, this is a good area to explore. And then when, then we came onto the other side of it, and when, then we tried to go through the capabilities of the service lawyers that have they done something which is different from the last year on the underwriting part. And what we heard from them is that a lot of them had started developing digital underwriting solutions. Now the adoption of those solutions is something that, that still, the verdict is still not out on that. It’s gonna still take time for all the enterprise to embrace those solutions. But, so these solutions started existing in the market. That’s where we thought that, okay, look, this is where we wanna write something about it. Now, when it comes to the, the key thought that the, the key pointers that we uncovered during this is that that underwriting itself is a complex process. Underwriters spend 60, 70% of their time on non-value activities. If we look at their current work profile and <crosstalk> did you say, sorry,
CW: Did you say 60 to 70% of their time?
AK: Yes, yes. 60 to 70% of their time on non-value activities. And when we look at the the amount that underwriter, the, the charges or what the enterprise have to spend on an underwriter, it goes anywhere between $50 to one $50 and R. That’s huge. Now, assuming that 60% of it is just getting wasted, that, that, that’s really something that companies should be thinking about it, especially in times where the claims the issues have been going bad for the insurance enterprises. They are taking a hit on their expense ratios and they’re taking a hit on that bottom line. And this is one area that they can easily go ahead and outsource because not a lot of outsourcing has happened on. Right? So that is where,
CW: I’m sorry, can I, I want to interject. Yeah, yeah, please. So 70% non-value add, but does that non-value add activity take? Is it like really specialized knowledge or training you have to have? Or is this sort of fungible activity?
AK: No, so, okay, so let me I was thinking of delving into it a little bit later, but since you have asked this question, let also put across some of the the, those activities that they have to, to take part of, right? And, and which takes a lot of the money. So the, the biggest activity, which takes their time is the initial submission intake process. The problem with this activities that what happens is that the, they, they deal with brokers day in and day out, and brokers usually end up asking for hundreds of documents from their clients, and they just dump it on the underwriters, irrespective of what the <inaudible> the case is being considered for what the documents are required. It’s just a dump that goes there. Now, underwriting has to go through these documents just to figure out that initial set of information, bare minimum information is required to even start the process. Now, that’s just the first step of it. The problem also happens is even though they’re sending a lot of the document, that doesn’t mean that they’re sending all the required documents. Then there’s a lot of back and forth, which also happens. Okay, this information is missing, this document is not more the most updated one. Can you send me the new version of it? Right? So all this initial processing itself takes a lot of days, and there’s back and forth with the customer, which keeps on happening, and this takes a lot of their time. Thank you.
MG: And I would say AB Abby, the, the next step, right, is, so all everything you described is the back and forth to just make sure that that submission is, is what I’ll call in good order, right? That you have everything you need to, to do analysis, but once you get it in, then you actually have to validate that the information is correct, right? And there’s a lot of data error potentially in some of those documents, whether it’s coming from the broker, whether it’s the, the end insurer that maybe submitted something incorrectly, or it doesn’t line up with all the fields that that underwriter to input correctly. So there’s the, the, the intake or the ingestion of, of everything you need. And then there’s the validation that everything you need is in good order to even begin the process of doing the underwriting work, right?
AK: Yeah, yeah, definitely. But what you’re stating is it’s a real pain point within the industry, right? So, so, so like you mentioned, the first step itself is where getting the those data points is a big problem. And then once that happens, even validating some of those data points, if the information which is being put across, whether it’s the right set of information, which is there or not, all these things, takes a lot of their time and and effort. And then ultimately what happens is that it also results in a poor customer experience on the other side of it mm-hmm. <Affirmative>. So while the, you have the most expensive resource in the industry dealing with this, this particular activity, which doesn’t even require them to be part of the process, and still at the end of it, you end up izing your customers because they have a very poor experience of the whole process. So, so that, that’s like the double whammy on the industry. They can have a better solution for it, which will not cause them as much and have a happy customer versus giving it to someone who is definitely not in for it. They didn’t ask for it.
CW: Yeah. That’s a big point. You were gonna take us down the road of talking about what the trends are and, and outsourcing. Yeah, it’s, circle back to that.
AK: Yeah. So I think, I mean, we’ll, we’ll circle back to these activities later. I mean, where all the, the, the digital underwriting can have a good amount of intervention and around it. But coming back to the, the trends which I was thinking about, right? So so from that perspective, we saw that okay, the, this is one activity that can be the be outsourced the most. There’s a potential for outsourcing. There’s a demand, there’s demand that is there, there’s a supply which is coming up. And the other factor which kind of helped during this process is the covid covid kind of accelerated this trend. There is <inaudible> that the talent wasn’t available. Underwriting, if you look at the current insurance industry, most of the tenured resources are to their retirement cycle. So what’s happening is, and the new talent which the industry wants, it’s not readily available.
So that’s where these enterprises thought that with the scare amount of talent that is available, we should rather deploy them to the core processes instead of you deploying on to these non-value activities. So that’s where they started thinking about outsourcing these activities to service providers. Now, on the other end, what happened is that these providers, while they developed these digital solutions, but they also started getting the right talent and the people with domain expertise to help build these, these digital underwriting solutions and also have operational resources to cover for the exceptions which are gonna be generated out the, out of this process. And that this is where this worked really well for the industry. And as far as what we have seen. So we, we started the process of writing this paper a year back, and we released it not year back as such, but couple of months back around that.
And we released it after that. And it’s been six months since we would’ve released this paper. Still, we are hearing from the market that more and more digital underwriting solutions are coming into the picture. In fact, the interesting thing is that some of the general use cases that I’m hearing, they are also targeted towards underwriting. So that’s an interesting thing that the one activity where there was literally no outsourcing happening before two, three years to where it is being considered a candidate for a generative use case. It’s, it’s just something that someone would’ve not imagined a couple of years back.
MG: Yeah. We’re probably gonna have you back next year to talk about how generative AI has impacted the underwriting workflow.
AK: Yeah, definitely.
MG: I mean the, and, and this may be taking us back to the point that you’re gonna make later, and if so please don’t let me divert you any further, but the question that that comes to mind is of the, the vendors that, that you were looking at are, are they solving for solving digital tools to, to the, I don’t wanna call it for the benefit, but for the underwriter themselves to use, or digital tools to streamline the process so that the underwriter can focus on different work? So meaning are they, are they outsourcing some of the process so that it can be a straight through underwriting workflow, freeing up the underwriter to do a different type of, or, or more specialized or product lines or something like that? Or are they digital tools that the underwriter is actively in that just has improved how efficient they can be in their,
AK: So, so Michelle, an interesting question, and my response to this is that both I’m seeing both in the market. So, so starting with first, right? Which is about streamlining the process. So that is where having a, a digitized workflow or bringing in some of these sub the submission ingestion tools, which, which works around extracting the, the documents and digitized fashion using O C R and I D P, and then using some of, now also using potential gen AI tools on top of it to make sure extraction is almost there. And there, there are very less amount of man manual exception errors which are being generated. And two, having some of those web portals for the customers and self-service tools so that they can track the overall process from the start itself. They don’t have to be dependent on the broker to understand the whole process.
So the, the, these, the initial steps where it, it was about solving the, the overall process. And then when it comes to triaging, there are triaging solutions also which came into picture. These are the, these three processes were being streamlined through digital tools. But the next set of tools that came in, which was around risk and pricing, those were enablers for underwriters. So while underwriters were taking, right now most of the decisions through the amount of experience that, and through their intuition and gut, what these risk, risk and pricing tools are enabling them is to take those decisions better. So this risk solutions which are available, they take into account so many different data sources, which underwriter didn’t have access to earlier. It uses the historical data to draw out the trends and then creates a risk profile of the customers of the potential customer using all that data set, which is available.
So that before even that case comes to the underwriter, underwriter has some clue of what sort of risk I’m looking into. Now, that doesn’t mean that underwriter is blindly taking the decision just on the risk profile that is being created. That’s just the step one to it. It is there then underwriter applies his logic and whatever rules that he internally has and takes an underwriting decision. But this is definitely a good enabler to start. Now, on the other end of it, we’re also seeing a lot of places that they straight through processing within the underwriting, which is happening where the underwriters are not involved. That is also one of the area which is more process driven, where what underwriters have done is that they have helped create those tools and those rule-based engines, which are the critical rules, which needs to be assessed for going through straight through processing of a particular case. And after defining those rules and going through multiple iterations, they have made the system so pool foolproof, that certain amount of coverages and certain kind of policies just goes through straight through Only the complex cases are where the underwriters have to deal with and take those underwriting decisions.
MG: Yeah. And are you saying,
CW: Yeah, go ahead, Michelle. We obviously have a lot of questions because they’re very fascinated by this. The, the report was, was great. I, I wonder from, from your perspective, and I know that within lines of business, there can be you know, different needs to, to your point, some of it is more straight through processing. Some of it is tools for the under. Taking it a step up just broadly between property and casualty and life and accident, are you seeing SL trends or areas of focus from, from your customers, or are they divergent a little bit just by nature of, of that type of insurance?
AK: I thought for, so interestingly, we talk about both life and annuities and the property and the casualty, right? So so the focus of the paper was purely around the commercial lines within the property and the casualty mm-hmm. <Affirmative>. But what, what we have been seeing is that the amount of trade through that’s happening across that differ. So if we, if we talk just about the personal and the commercial lines within the property and the casualty, personal is where most of the street through processing has started happening. Yeah. Commercial is where the intent is to go through straight through processing, at least for some of the cases. But right now it is being done more manually. And when we look at specialty lines within the property and casualty, they, that is where the scope is, the minimal at this point of time because of the nature of the the risk that comes into picture are over there. Now on
CW: The other, so that, that scope, sorry, for the specialty lines, is the scope then sort of restricted to the earliest parts of the process, the data entry? Or has it creeped past that?
AK: No, I think it, it is, it has more to do with the data and only the, the reason being that we are talking about the, the cyber insurance. So we are talking about D N O or we are talking about marine. There are not enough, even for the the service providers or any tech enabler to develop a solution around it. They need a robust amount of data to create those risk models. The problem is that that data doesn’t exist is at this point of time, you get these risk once in a while, right? And even the nature of the risk keeps on changing over the overnight. So that’s where developing a digital solution for this segment of the market is going to be a little bit more difficult. It, it might happen somewhere down there in future when there’s good amount of data to back it up, but at this point of time, it’s, it’s one of the most difficult thing to achieve.
CW: Yep. That makes sense. Sorry, continue your train of thought. Yeah, I interrupted again. Yeah,
AK: So, so that, that’s on the personal and the commercial lines. Now, when, when it come to the other side of it, which is the life and the annuity here, the amount of state through processing is definitely on the higher side, higher when it come when compared to P N C. The reason being that the, the, if you look at the nature of the businesses there around life and annuity, a lot of things are, are predictable in a way. And lot of you can train an underwriter engine to look at these certain parameters, which underwriters are looking right now, and create a risk profile around the customer. You have certain list of questions that, that you, that as, that are part of those insurance form and that are being asked by your customer. And you just take that into account. Along with that, all the, the past claims related data, it’s medical data that’s available and all of it. You just combine all of it and you can easily create a risk profile. So the amount of straight through processing over there is definitely on the higher side, right? Compared to the, when we look at the pnc. And I think that that’s gonna be trend going forward in the future. Life and ity is always going to surpass the PNC when it comes to automation, especially on the underwriting.
CW: Yeah, that, that makes a ton of sense. So with all of that as background, what do you see as sort of, what are the trends that are just starting to emerge? Like where, especially with commercial lines and property and casualty, where do you see those industries headed in terms of automation in the near term?
AK: Yeah, so I, I think I like to break it down by the personal commercial and the specialty, right? The reason being all the, the nature of the business is fundamentally different, all that. So, so like I mentioned earlier, in personal the straight through processing is completely penetrated in the market right now. If you look at it, it, in most cases those, the, for, for the personal lines, the digital solutions which are available in the market, they have, they have been implemented in some shape and form. So that is where the market is looking for more new next gen e i kind of solutions. Okay. And okay, what additional you can bring in into, apart from some of it has already been implemented. So what, what is the next for the industry? So they’re, they’re talking more about, okay, how can, so, so this is what has happened.
There are, there’s already a digital underwriting solution which is available, but now even I wanna enable my underwriter even more for those complex cases that underwriter is considering. So, so is there a potential Chad g P t kind of a solution which can, where underwriter can ask a question and get a response based on all the hundreds of data that is set, that are available, and instead of scraping through that data and taking like 40, 50 minutes on a single case. So that’s where the industry is looking forward to when it comes to commercial line, the industry still learning about it and implementing the digital underwriting solutions. So that is where, obviously, because it’s a new flavor in the market, so everyone ask once in a while that, okay, is there some gen AI component in it or not? But when it comes to implementation, they’re still looking at more rudimentary solutions, which are available in the market. Yeah,
And I think specialties where if you look at <inaudible>, they’re gonna be far behind personal and the commercial lines. The reason being that while they also are looking at various technologies to enable their business, but it’s still going, it’s still about figuring out the, the fundamentals of the business. First, if you talk about cyber, the cyber insurance, while there are a lot of cyber claims that has happened in the last few years, the problem is that the nature of these claims keeps on changing. Even the, the companies which are into these they don’t have ability to launch new products, design new products, or go ahead and underwrite some of those risks because risk keeps on changing. So that is where their reliability is more on their underwriters to help them versus these, some of these commercially available solutions, because neither the adoption nor the competency of some of these tools are at par with personal in the commercial lines.
MG: And I, I think too, Abby, right, the, for those specialty lines where the, the process is in need of, or the goal is to automate some of the, the data intake, right? Or hit those third party systems to, to bring in data automatically, those, those data sources aren’t as fully baked or robust as some of those that are used more broadly in commercial or personal lines. So you’re still, they’re still gaps in the data, even if you are trying to automate that process. Whereas to your point in per in personal and, and broad commercial, you’ve got, you know, public and private data sets that can be purchased that are pretty, pretty close to being complete, I’d say.
AK: Yeah, definitely you are, right. The, the, the biggest problem this in, so insurances as an industry is a risk covers industry. So the problem is that, that anywhere they, they end up seeing good risk in any of the activities that they’re attaining, they will take a back step to it. So right now with, there are, there are still some public available data sources on these specialty lines, but the problem that the, with those data sources are, they’re not complete. They and industry doesn’t trust those data sources. So that’s why that’s where they end up taking a back step into it. But the ones which are available for the portion and the commercial lines they are tried and tested. They have been used by the industry long before all of this came into picture. So that is where the reliability comes into picture. And they are okay with trusting some of those models
MG: For sure. Yeah, that’s, that’s why it’s, you know, for startups coming in, in, in my world, it’s both intriguing to see where, what is these, these new data sets, how are you acquiring that information? But then it is also a challenge when you’re up against some of these incumbents vendors that you may actually be supplemental and hopefully not a hundred percent competitive. ’cause To your point, it’s a challenge to, to break through <laugh> and, and beat out one of the, the big players.
AK: No, no, definitely. And I think what keeps on happening is that as soon as some of these big players, they see that, okay there’s a startup or an InsureTech, which doing good, they’ll end up either partnering with them or acquiring them instead of letting them grow. So also, that’s it. And the ones that are not picking in the market, they’re anyways getting thrown out of the market. Especially now since the valuations have dropped off some of these startups and InsureTechs, they are even struggling to get business and survive. So it’s, it’s all a game of survival now for these startups. And the only way that some of these startups and InsureTechs can move ahead is by partnering with these large players. They can’t go ahead and create a market on silos, the reasoning that they don’t have the capability or the experience to handle it.
CW: Interesting. We talked a lot about digital underwriting, and I just wanna sort of widen the view a little bit. Are you seeing, or anybody that you, you know, your colleagues at Everest, are you seeing similar trends in claims?
AK: Yeah. So on the claims side, obviously claims had been digitized much before then the, the overall underwriting into the picture. So what we are seeing on the claims side is that, that initially it, it was all about the, the, the admin work, which was around claims, right? A lot of it has been outsourced, and like I mentioned earlier, it has been saturated. But now the, the kind of asks which are coming on that side, especially on the automation front, is that whether you can identify the claims leakage for me. So what we are seeing interesting trend that we are seeing in the industry is that service providers are putting the skin in the game. They are saying that you give me x y you gimme me, this whole what of whatever claims outsourcing that you can, to me, you gimme end-to-end process to outsource.
And what, in a way I’ll help you do is that I’ll help you identify claims leakage, and I don’t. And the deal is structured in such a way that instead of service providers being charged upfront about being charging upfront about the resources and all, they’re going to gain share kind of construct where out of the percentage of the savings that they’re able to generate for the insurance enterprise, they take a cutoff. So that’s where what we are seeing is that, that with the, the, and, and this, this, this hasn’t happened overnight. It’s not like there’s some clean <inaudible> solution, which is just, which was just available with the service product. And they, they came with a magic wand and the just healthy enterprises, it has gone through the journey. So it started with getting those data into order. The first step was making sure that the data is available, that the data is structured, that the data is flowing through the systems and all it’s getting captured in the right way.
And right now they’ve reached to a stage where they have a hundred percent visibility of the data. Using all of that data, they are confident that they can go ahead and stop some of those leakages, which have been happening. So that’s an interesting trend that we are seeing over here, here. And like I said, the gen ai solutions are being developed on the underwriting side. Similarly, we’re seeing trends on the claim side where using some of this dataset and the Gen AI capabilities the leak, the, the main intention is to prevent as much leakage as possible.
CW: Yep, that makes sense. I want to, I, I think I’ll probably get fired if I don’t ask a direct gen AI question, but I don’t want to, I don’t wanna go there too, too soon. So Michelle, anything else you wanted to do to tie a bow on this topic?
MG: Just that, I think it will be interesting to track track Abby, both the, the continued insights on the underwriting reports that come out, but also on the claims because the other item that we talk a lot about is how they’re, they’re connected, right? That the more you understand about underwriting, the better your claims, you know, should be. And then the, the more insight you have on the claims, that should inform your underwriting and hopefully streamline and help automate some of those processes on the front end. So you know, I recognize the two separate groups within Everest working on that, but it’ll be interesting to see if those trends continue to align in kind of a similar fashion.
AK: So, so it it, it’s the same team which keeps a track of both the, the claims and the underwriter, whatever digitization that’s happening on the claims front. And, and, and you are a hundred percent correct, right? I mean, the, so, so, so it’s kind of a a cycle where, where you understand the first, then you are better able to predict this, the, the other one, right? So initially it all started with the claims and the, the reason that right now service providers are confident about some of their underwriting solutions is because they have a very good handle over claims data. Mm-Hmm. <affirmative>, they’ve worked through this claims data across years, they understand the nature of the business, they have developed their domain expertise around that claims data. And then, and they’re using that to, to nurture their underwriting talent, that service provider’s underwriting talent, which is there, and also develop some of these models.
And this, this synergistic capabilities between the two is also being utilized by the enterprises. So when, when you look at the underwriters helping design some of those rule engines, a lot of this dataset comes from the actual claims data which is there that, because that, that there was a claim which happened somewhere, was someone was not able to predict that better earlier. So that came out as a call out in the new set of data, which was coming through that, that, okay, just taking an example of cyber insurance, right? New kind of cyber insurance for the real claims started coming in. So that’s where industry started realizing the current products are not relevant for the industry. They to design new products. And how do you do it by using that claims data? That’s just one example of it, right? If even if you look at the personal and the commercial lines there also, we saw that the lot of these claims exposure was coming through.
This helped take decisions for new product launch for developing some of those risk and underwriting models, which were there proper pricing them as well, right? You or interestingly enterprise that using the claims data to understand that, okay initially it was one size fits all approach, but now they don’t wanna wanna do it. They wanna have a customized approach for each of this, their customers. So that’s, how do you do be beyond the public availability data, which is already there? What is the next thing that you can utilize to for a customized offering? It’s obviously the claims data that you have apart from the other data set about the customer and all, but claims is, is an actual predictor of, okay, how, how this particular policy that I’m issuing is going to perform in the future, right? So that’s where the, the point that you mentioned makes sense. A hundred percent.
CW: Great. All right. And now the, the big question <laugh>, pull out the crystal wall in insurance where, you know, where you spend your days. One, do you think we’re close to or past the peak of the gen AI hype cycle? And two, what use cases or applications of gen AI in these industries that you’re an expert in, do you think are gonna stick?
AK: We are, we are nowhere past the hype cycle. We are just starting to be honest. We are, we are eight months into it from the time the this gen AI evolution started. Yeah. But it just feels like we are the starting, we are just seeing the tip of the iceberg right now. We, we still have to encounter a lot. So right now, when we look at the industry, what’s happening is that there was a mad rush for, within the industry to develop the generics use cases because everyone, everyone knew that everyone’s gonna ask for it. Most of the insurance enterprise, even if they are at a space where they can’t even implement ai, but they have started asking about gen AI solutions, right? So, so even if the, so it’s a, it’s, it’s a must have solution as part of the portfolio. So that’s where, what happened within the industry that most of the service providers started realizing that, that, okay, I need to do something about it.
In interestingly, insurance, what everyone knows that insurance wasn’t the first first industry where the J AI application started. It all start, if you look at the P P O industry, it all started with the customer experience side contact center. That’s where the first use cases started getting developed. But then once these large language models, people started realizing the potential of them. That’s where the insurance industry also started developing some of those use cases in areas like underwriting claims, policy servicing and all. And they started doing a P O C with the client. So, so interestingly, what, what they’re doing right now is that they, they don’t have full blown solutions available that they can commercially deploy multiple clients. What they’re doing is that they, they develop a use case, they take it to a client when client agrees to it, they do a P O C for a portion of that business.
If that succeeds, then they get, then the idea is to implement it to across a larger business and take it to other customers. So through to what’s happening through this approach is that they’re able to test multiple ideas at the same time instead of putting all their eggs in the, in the one basket where they end up developing a solution and then a competition beats them to it with a better solution or with a better use case. They’re, they’re taking a different approach. They are, they’re looking at all the possible use cases, which can be developed, but they’re prioritizing the ones where there are most interest, which are being generated from the enterprise side. So that, that’s a novel approach that they’re taking to genea. And I think when, when we look at it from the Everest perspective, also, we also think that that’s the right approach to go with instead of just developing a solution. But because usually developing a full blown solution is gonna take some amount of time and you don’t wanna miss it if you end up developing a wrong solution at this stage. No. So I think but that, that’s, that’s more on the approach part of it. That’s what we are seeing in the industry now. The venue mentioned about the, some of the use cases which I’m seeing in the market. Yeah. What
CW: Do you think is gonna stick?
AK: So, so to be honest, it’s too early to see, and I think I like to use it depends right now. There you go. Which Michelle said earlier,
CW: Michelle gets $5 every time someone says that on the podcast. I
AK: Encourage it. <Laugh>, <laugh>. So so coming back to, to the, to the topic, right? So I think more than the, the what solution is gonna take market, I think what we should look at is which are the solution which are gonna get implemented first and what’s gonna next come in line? Because it’s, it’s not a revolution that’s gonna stop. I think Metaverse was something where initially it was similar hype, what it is right now for, but it didn’t stuck around for long because there weren’t made that many possible number of use cases that were there for the industry. But gene AI is used, like people have been using the enterprises and the service products have been using AI from so long. It’s a na natural extension of the those and that, that particular capability and bringing in gene AI capabilities into it, right?
So it’s not going to go down as a hype anytime soon. So the initial set of use cases, and these are just the preliminary ones that we are seeing in the market. Some of it is around underwriter, underwriter think process, which is about underwriting assist, where, like I mentioned earlier, that underwriter gets a chat, G p t kind of an interface where for any of the keys that underwriter is working on, underwriter can type in the tho those questions in that particular solution and they can end up getting to it instead of scraping through those 50 documents, which are there, this underwriting solution, gen solution going to extract information from all the, the data that’s gonna get captured through brokers, and it’s aiding the underwriters in taking that decision. Now, how it’s going to help underwriter is that right now, if it, if we look at any of the cases, which underwriter is looking at it, underwriter typically spends around 50 minutes to 60 minutes in assessing that particular case.
Now, e even if they are these underwriting solution, the generic solution that’s available in the market is able to reduce that time by 50%, that’s a huge win for the underwriter. So that, that, that’s one on the underwriting side. The second that we are seeing is around the claim side, which I mentioned, which is around stopping the potential claims leakage, which is where they are ingesting lot of these claims data signals, which are coming from multitude of sources. And they’re predicting that, okay, which, which kind of claims are the ones where the potential of leakage, what are the initial step itself that you can enable the, the claims agent to take into, so that leakage can be stopped at that point instead of system detecting it later on. So that’s more on the claim side. There are also some of the other solutions which are around generating insights from the large data set, which is there, which helps the enterprise at the top level.
So le let’s say C F O gets a hundred percent visibility of the data. So what, what we are interestingly seeing is that the, with these models, the C F O can just write that, okay, I wanna the, the summary of my sales of x, y, Z product in the last four years, it’s gonna automatically generate the graph and the insights around it. Yep. and these solutions are, are increasingly being pitched to the top management because that is where the use of this solutions are, are much higher. I think these are some of the areas
MG: We’ve, we’ve seen a number of, of those use cases or solutions coming up as well. Just basically how do you interact with your data? Yeah, yeah. Right in, in an easy way. And typically these, what you’re talking about obvious is they’re being geared towards business users, not data users, right? And so that’s, that’s I think a trend that we’re starting to see in terms of the value proposition that some of these companies are trying to use generative AI to solve for.
AK: Yeah, definitely. And I think if, if you look at the key decision makers also, they are not the, those data users, they are the business users, right? And, and gen is something that’s a, that has become a C X O agenda right now, or someone at the top is driving that agenda instead of a business user or going ahead and saying that I want a G N A A solution in my day-to-day operations. Can, can I please get one? Right? So that is where targeting them makes much more sense versus targeting some of those data users because you might not get that high adoption if you target this limited set of those users.
CW: Absolutely. Well, this has been a fascinating conversation. You’ve been listening to another episode of Unstructured Unlocked with my co-host Michelle Govea and our extremely engaging and unbelievably knowledgeable guest, Abby Koi practice director at Everest Group. Abby, thank you.
AK: Tthank you so much, Chris. And Michelle, I think it, it had been a learning experience for me as well. I think lo knowing from you that what you also have been hearing and seeing in the market and what’s happening, so I think definitely we are gonna use some of it in the <laugh> the next set of reports that we’re gonna publish, and we’ll definitely be in touch. If there’s something that that’s there, I’ll reach out to you and I’ll just get your perspective on it. And I, I’m sure you will have other hosts to ask some of these questions too, so I’ll get a perspective from you
CW: Looking forward to it. Yeah, thanks again.
Check out the full Unstructured Unlocked podcast on your favorite platform, including: