Actionable Insights From Unstructured Data

INTERVIEW

On episode 35 of the InsurTech Geek podcast, talking about actionable insights from unstructured data with Sai Raman from CogniSure. 

The InsurTech Geek podcast, powered by JBKnowledge, is all about technology that is transforming and disrupting the insurance world. We will be interviewing the minds behind the innovations that are driving technology in InsureTech. And doing deep dives into specific technologies that we see changing the industry. We are taking you on a journey through insurance tech, so enjoy the ride and geek out! 

JAMES: Alright. Another week in the 4th quarter of 2020, Rob Galbraith. The 4th quarter. This nutty year is one quarter away from being done. I do not know about you. But I am not going to shed a tear when we cross into 2021.  

ROB: No, I think we are all ready for this to be over. I do not know, James, if I have told you about our political yard sign that we have at my house, it is not for President Trump, it is not for Vice President Biden. It actually says giant meteor 2020 – just ended already. 

JAMES: Exactly. That is the general sentiment this week in particular act of the presidential debate is, let us just get a meteor and finish this off. It is the dumpster fire heaped on a dumpster fire. It is a crazy year. And again, we try to stay super apolitical on the show in general. And of course, there is nothing more apolitical than a giant meteor, so I do like that, Rob and I agree. It has been a weird year. I think the only saving grace has been that we have had college football back towards the end. I went to the very first season opener at Texas A&M. They played Vanderbilt and we have managed to barely win. And you could tell how distracted the players were cause it was just a weird game, weird vibe, but an Aggie victory made it taste a little bit better. And with us today, Sai Raman. Sai, it is good to see you. Where are you joining us from today Sai? 

SAI: Chicago, James, how are you?  

JAMES: I am doing really well. I love Chicago in particular in the summer and the early fall. I am a Cubs fan and I love my windy city, so I am glad you are joining us from there. It is a beautiful time to be out there in Chicago. isn’t it?  

SAI: Yeah. I think it is getting cold, but Chicago in summer is you know, you cannot ask for more.  

JAMES: Yeah, exactly. It is a really nice up there. So glad you are joining us today. Before we get started with our interview, do not forget, you can subscribe to the InsurTech Geek podcast by texting GeekOut to 66866. Make sure you never miss an episode. And now back to our special guest, Sai Raman from CogniSure. Sai, we are going to talk about CogniSure, we are going to talk about just technology in general and what AI really means. What portion of AI you are using, cause, no one’s created general AI that we know of yet, right? We are all dabbling with specific forms of artificial intelligence, but I want to talk about you for a second.  

You got your Bachelor’s in Andhra University in India, and then you went to Madras for your Master’s in Engineering. And then you worked at a bunch of companies I am super familiar with. You worked with some very large, Indian outsourcing and consulting firms, Tata, Cognizant, you worked for several years over at HCL. You obviously have had a bent towards engineering from a very early age cause you majored in it and then you started doing engineering and technology practice consulting. So, what was it about that that really attracted you when you were a young kid growing up in India? What made you want to be in engineering and go work in technology?  

SAI: When we were growing up, the word technology did not exist. It was engineering or medicine. We had two choices. Engineering was cream. Getting into engineering was really tough, it was a challenge. So, I wanted to take the challenge at least and get an engineering degree.  

JAMES: And once you got in, you jumped straight into a consultant role with Tata. They are a massive company now. They are just enormous. What practice area were you in there? 

SAI: Yeah, it is interesting, right? Actually, in India from 1996 to 2000 was that Labor Arbitrage rush, because in the US, that is when the actual Indian IT got the moment you can call the tipping point, right? A lot of people like me who never thought about information technology, got the experience like the startups and the Infosys of the world started and we were all excited. Mainly you could travel abroad. You could work on multiple projects. So my major was mechanical engineering. In IT Madras, I was researching welding technology. My specialization is welding technology and engineering. And then I got an opportunity to be in TCS as a software engineer. And that is the way I started to be IT. 

JAMES: Well, that is a pretty big difference. So, you spent time studying orbital welding and TIG, welding, and you are like, okay, let us just jump into software.  

SAI: Exactly. Yes, it is interesting. When I jumped into the software, I came to the US in 1997 and my first project was working for Target Corporation. We were supporting their systems and then I went back to India and then they made me a project manager for an insurance project. So, I always tell the story of how I became an insurance person not by choice, but it just happened. I was the project manager to integrate two company systems. I had a 30-member team, and I had no idea what we were doing. I have hired 30 IT people, so we were migrating policies and contracts and claims from one system to another. And we used to call the file names Prem-sum, Pol-sum. Six months into the project, we realized, oh, Pol-sum means it is policy summary for them. Prem-sum means it is a premium summary, so it was such a mess. You, as a project leader, you do not know what you are doing. And I thought, okay, I need to understand insurance. And that was the beginning of my insurance career. I would say 1998 or 1999.  

JAMES: Yeah, it is how many of us in technology got into in any industry we worked in. I jumped right into running a software company right after my senior year in college. I had no particular industry expertise. I had only done two internships with PricewaterhouseCoopers, so I had six months with a big consultancy. And they on some very interesting projects that I worked on. But I kind of knew what that was about but had to kind of figure it out. And then I ended up in insurance, three years after I opened the company. And it is the same thing. Like what is a policy, what is premium? I got two business degrees. I did not get engineering degrees. I got an Undergraduate in Accounting, a Master’s in Information Systems. They taught us nothing about insurance in the business school. I did not know what policy and premium and claims and all of these things were.  

And it is funny cause you literally interpolate it from field names and table names and your SQL queries. And you are like, oh, okay. So, it is a totally different way of learning the industry cause it is like learning how a body works by sticking a microscope inside someone’s body. You are looking at it like, oh, okay. I see how this all works. It is a wild way to learn a business, but certainly, you learn it pretty quickly, cause you have to learn how the data flows through an organization and that is a big challenge.  

SAI: Yeah, those days in India, if you are a technology person, you are from one project to another. One day you are in manufacturing, the other day in retail because you are a developer. Like Coldwell developer or Java. The first thing I did when I went to the UK, actually as part of the project, I went to the UK, and then I took the certificate in financial planning. I am actually a financial advisor in the UK. I thought that gave me all the financial products. I was the first person in my area to get that certification. Those days, domains did not exist. Everybody was talking about technology when nobody was talking about domains. When I started talking about it, they made me a domain expert and that helped me stick to insurance for the next 20 years. So, I am very glad that happened actually. That is the way, otherwise, you are not going to learn insurance just by doing projects. You need to make a conscious effort to really learn. So, when I came to the US after that, I took the CPCU. I did my accounting and finance in insurance. So, learning insurance does not come just by doing projects, I think you have to put a lot of effort to really understand the industry deeper and that is what I did. 

JAMES: Yeah. That is awesome, Rob? 

ROB: Yeah, that is nice. So, James mentioned some of the big companies that you have been a part of. But now you are the founder and CEO of CogniSure, an InsureTech startup. So, I definitely appreciate you know spending almost 20 years working at USAA, working at a really large company, and then going out on my own, like the difference is, so yeah, so tell us a little bit more about what made you start CogniSure? What was the business opportunity that you saw? And then I know, you are focused specifically on helping, loss run automation and standardization of an early use case. So, maybe you can share why that is one that you began your focus on for your company.  

SAI: Sure. As I was working with all the Indian companies. James, you were also asking about my outsourcing experience, right? India has gone through labor arbitrage. So now the world started thinking about the data machine arbitrage, it is more digital tools. A couple of years back, I wanted to discover a problem in the industry, which was never solved and was worth solving. So, my initial thought was, as an individual, as a homeowner, I wanted cover for all the risks which I was exposed to. So then, how do you find out if there are any coverage gaps in my own scenario? And the only way you can see is to look at your policy. And then somebody says, okay, you do not have this endorsement, you do not have that endorsement. I thought if I can come up with a way where somebody can say look, these are your coverage gaps, these are the gaps in your protection and the probability of these risks. It will be great for me as a consumer, but it is not there. I cannot check what coverage gaps I have in my policies.  

And you live in Texas and you know, 70% to 80% of flood claims people do not even know that it is a separate policy. Especially in Illinois, the water backup endorsements in basements, so my initial thought was, how do I find out coverage gaps in the policies? Or what is a production score, similar to a credit score? What is my production score? Where are my gaps? So then immediately you are dealing with PDF policies. What I have is my homeowner‘s policy, which is a PDF. I wanted to read the policy and see, do I have this endorsement; do I have this limit? So that started me on a journey of, can I actually get what I want from these documents? Because they are non-structured, unstamped, nonstandard unstructured. So that is when I started as an experiment. And I found out that I was able to solve that. It is not easy.  

You need to understand a lot of contextual knowledge of the contracts and policies. Once I figured out a way to do that, bingo, then I said, okay, then there is a lot of potentials. Then I started looking at and talking to BrokerTech Ventures, rather than going to the personal lines initially. Why don’t they focus on the commercial side? And then when I looked at the commercial side, the biggest problem in commercial lines is the lost ones. Nobody has touched it because it is extremely complex, and it is a massive one. So, my starting point is, is this complex enough? Nobody is there. And I want to try solving it. And that is the way I started with loss runs. 

JAMES: That is great. Let us talk about CogniSure for a second. Just give me the 30–second elevator pitch. If you had to explain to someone very briefly what it does, help me understand. 

SAI: CogniSure is basically actionable insight from unstructured data. Unstructured data in the context of insurance being policies, loss runs, score submission emails. 

JAMES: So basically; all the data? 

SAI: A lot of companies are focusing on extracting the data. I see it is a part of the problem. You extract it, then you need to understand what it is so that you can act upon the data. My goal is CogniSure is just not extracting, it is rationalizing, normalizing into a common industry format, and giving you actionable insights into the data. If you give me a hundred-page loss run, okay, so if I give you a GSN, so what good is it to a business user? It is no good, so I need to put that into context saying that, okay your workers’ comp claims because of these causes are high, so you need to focus on… See it has to be converted into actionable insights, so CogniSure is a company that can convert the data and then normalize it and give you actionable insights.  

JAMES: Awesome. And look, when you talk about unstructured data, that is why I said that is basically all the data because there’s so much unstructured data in this business. I mean, the policy is a contract, right? And it is a lengthy contract, and it is complicated. And then all your claim files, you have tons of bills come in and then you have tons of journal entries and diary entries, and transcripts of phone calls. It is just mind–boggling how much–unstructured data there are. If you have a claim file there’s photos and videos, you can really dive deep just in the policy itself, the policy, and all the endorsements, and then, there’s just structured data heaped and unstructured data. This entire industry from the very beginning of it, thousands of years ago has been built on contracts and documents, but not contracts and documents that were in standard forms, right? 

SAI: That is right. 

JAMES: That is really where the rubber hits the road in insurance is all the little details and the gotchas that you have to read, or that hopefully, your broker is reading when you are buying insurance, it is really complicated. We have seen examples of the financial service industry like there is an example of J.P.Morgan says they saved 360,000 hours a year by applying machine learning, and some specific forms of AI, reviewing loan documents. And what they are looking for is, basically, there is a lot of menial work in the financial services industry in both banking and insurance of people just looking for a very specific language. I know that CogniSure is not that simplistic right, where you are just looking for specific language and how it is structured. You are actually trying to develop trends. Are you trying to get into reserve recommendations and policy recommendations? Are you trying to do full policy review documents? What is the end deliverable look like at the end of the day?  

SAI: I have a very large plan, but you have to go through smaller steps. So my initial focus is like, for example, if you say policy checking, to your point, the goal of policy checking is to do what, to be able to identify what coverage gaps you have, whether all the limits are right so you have to be able to do that. Whereas when it comes to loss runs, you have a big construction company, when you look at the history of a company, what do you tell them? You have a hundred-page loss run from three different companies. Workers’ comp, general liability, and property. Okay. I now have three–loss runs and hundred paid, so what do I do with that? You want to know, oh, do these three things so that you can reduce your losses, or you can reduce your premium. What I am trying to do is rather than become an extraction as a service, I want to take one domain initially. Okay.  

Let us take the loss runs as an example. Extract any claim, whether it is commercial auto property workers’ comp, and then contextualize to make meaning out of it, then I will go to the next one. Then once I have the losses taken care of, now I will go to the policy. I can go to the courts. If you are trying to do every document and any document, I can say, you are going to fail. You have to pick up one. And my battle, I picked up is loss runs. So let me focus on loss runs because there are millions of loss runs are floating around and nobody has a solution yet. That is my focus right now, but I will launch into a lot of other things. To your point, there is a community aspect also. Let us say if I have more and more loss runs, I can tell you let us say this class code or this industry, on an average, this many claims, but why are you having more claims? Today there are no centralized claims. And if I succeed in aggregating a lot of claims data in commercial lines, I can give more powerful insights, which does not exist today.  

JAMES: And what is the end benefit result for the company? Is a carrier buying this, is a TPA buying this, is a broker buying this? And if so, does this make them more profitable? Does this actually drive lower staff time?  

SAI: Brokers are buying this because brokers, for example, if you take their largest customer, they have, anywhere between three to four policies or three to four–loss runs and it can be 50 to 100 pages. They do not have a way to see all the loss data in one in one shape, so they can advise them on loss prevention, reducing the premiums, and actually guiding the customer about the risk mitigation strategies. So today for them to understand the historic loss data, they do not have tools. That is why brokers are super excited. The second one is underwriters are looking at, see the brokers is again sending the same loss runs to the underwriter. The underwriter is trying to see if they can issue the policy or do they need to decline this coverage because this guy is too risky. So they have to spend so much time again, reviewing these loss runs. Underwriters are buying, in terms of reviewing the submissions and wholesalers. So the entire industry risk mitigation, loss control, loss prevention, everybody has a view into your historic data of claims so that you can actually make a lot of meaningful decisions like reducing losses and preventing all that. 

JAMES: Awesome. Rob, I know you have a question about brokers.  

ROB: Yeah, so you mentioned being part of the initial cohort for BrokerTech Ventures. So just kind of curious a little bit about your experience. And obviously, a lot of carriers have been trying to partner with startups, and I know you are exploring those on the carrier side as well. But what attracted you particularly to work on the broker side and to focus there and then again, I would love to hear more about BrokerTech Ventures. 

SAI: My whole life in 20 years, I think mostly I spent with the carrier side and especially the policy admin claims billing. The last two, three years, I got a chance to work with brokers and I said, oh my gosh, I think brokers have not invested heavily in the technology. And the other thing is, the industry has always focused on the product. I am selling this product. Nobody’s thinking from a customer standpoint, I am seeing the customer standpoint. Okay. This company is selling an auto product or worker’s comp product, so BrokerTech Ventures, I thought is the best way to understand the customer. Because the broker is trying to help the customer and that priority or that focus is missing in all the digital initiatives of the carriers. They are more interested in, how do I underwrite better, how do I drive system efficiency? So, BrokerTech is a channel to understand the customer better because it is close to the customer. So I said, okay, I am going there. And then when I showed him the loss run, he is looking from his customer’s eyes.  

What can I tell him, tomorrow, why his the premium is going up? I should be able to look at and say, yeah, you have to control your claims. So, I think BTV is a very powerful accelerator there in the market. Number one. And because they are not too large. Their investment and time are significant. And they are really working on pilots. I do see they want you to be successful and they are working, and it is different. A lot of involvement. You learn a lot, I really enjoyed being part of BTV and there is a tremendous value, and you see a different angle if you are in insurance, and mostly you are from a carrier side, but you go to a broker, you get a different perspective of a product or everything. So that is what is very important. And I really liked it.  

JAMES: Okay, let us talk about what is next. What is on the horizon because right now what we are doing, I feel like we are in the infant/toddler phase of AI, right? We are learning how to use machine learning and software robots. We are learning how to teach machines, how to learn, we are letting them learn, and we are using the output to automate a lot of really boring, menial tasks we do not want to do anymore, and to identify trends that we do not have the time to analyze, right? We do not have the budget available to pay enough people to sit and peel through unstructured data, to get any information out of it, which means we have gotten really untapped exabytes of data that have never been analyzed before. One of the things that really excites me honestly is the potential to go back and back scan, historical documents, or take, and I have been watching some other machine learning and AI companies doing this.  

They are taking a document repositories, really old case files, old claims files, old records, going back 10 and 20 years, the ones that have already been scanned in and they are using natural language processing and they are ripping them out. They are using OCR technology and then they are actually analyzing all the notes and then they are correlating the notes to outcomes. And then they are looking at old policy documents. So, there is a lot of work going on right now. There are hundreds of millions of dollars, if not billions of dollars, really chasing some really big problems. And so that excites me, cause what is going to come up is we are going to learn how to identify key drivers in the cost of risk that we really have not been able to identify yet. So, knowing all of that, what is on the horizon for you? Cause we have talked about where you are at right now and what you are doing right now, but what is next?  

SAI: I think what is missing, in my 20 years of insurance experience is the customer view. So, I always ask a question. If you go to a business owner and ask, are you covered, number one, for anything that can happen to you, are you covered? Or do you know what you are not covered for? I am thinking, this unstructured data, if it can make a difference to the end customer. There are industries who we will spend, underwriters will spend a lot of effort, but what is the purpose of insurances and the shifting from indemnification to prevention? Can you prevent the losses from happening? I would say, what is next for AI is, it is just all about preventing the losses from happening rather than, how do you have all this unstructured data?  

So, I would say the more focus is going to be shifting from indemnification to prevention, and technologies like, if I know, for example, I kind of myself saw that okay, I am in a zone where there is a risk earth quick or mining risk, I do not even know that. So basically if I know, what can I do about that? What kind of coverage should I have? So what I’m trying to say is more, the insurance has to start focusing on more the impact on the consumer because the carriers are always thinking about, okay, I implement the Guidewire platform, so it’s not going to make a difference to me as an end customer. So I would say the next few years, it should be, like even in your construction. I have seen how the construction industry can benefit a lot. 

JAMES: Oh, it is amazing. So many things. One of the most inefficient industries on the planet. People in insurance say that they are inefficient, not even remotely close to construction. Construction has razor–thin margins for a reason. One of the only industries on the planet that over the last 40 years, has not uptake at all in productivity output versus manufacturing. Everybody else has increased productivity, not construction. There is a ton of money chasing efficiency over in construction right now as well.  

SAI: Yeah, no, because even in the insurance companies have an interest. What I see is, insurance companies right now, rather than I read a policy versus what is future to me is, how do insurance companies collaborate with the end customers to prevent losses and make them better, rather than, I have a policy, I am getting a premium and you go and do whatever you want. Because you are absolutely right. I had a special interest in construction. And when I looked at a lot of research, as I mentioned, your report was amazing. JBKnowledge ConTech Report was the only report I could find, a lot of statistics around constructing you know with less than 1% spend. See insurance can drive the construction industry. Okay, if you implement this technology, it is, win-win. You benefit and as well as me, because the highest claims happened in the construction industry, so I think to your question, the collaboration is going to be shifting to prevention, collaborating with the insureds to make a different ecosystem is what I am hoping for, but I do not know when that is going to happen. .  

JAMES: On that note, Rob?  

ROB: So Sai, I’m really interested, having seen your software and seeing how it works and your capacity today to take loss information and not just take it from a PDF and structure it and create, filling all the data and summarize it and be able to identify the causes of loss and things like that. But you have dashboards, you have analytics, you can boil all that up to kind of a one-page summary. You mentioned an underwriter before, but from our broker side, to help advise our clients, and from an underwriting side to say, is this a risk I want to accept or not? Possibly applications and loss control as well. So I kind have of a two-part question here.  

Number one, maybe you can talk about that added part. Cause the part to me that is really new is it is not just going from the unstructured to structure, but you are actually creating the intelligence, right? A lot of maybe carriers can do that, or they have gotten data science teams or whatnot, but you are making it easier, for your customers to be able to gain those insights. So, what extra steps do you have to do to basically create the analytics? And then I am curious, as we talk about unstructured data in general and how much there is insurance, are you a purely going to be focusing on documents? You talked about policy language earlier, or are you thinking about, images, video? Are there other things that are on your roadmap that you would like to do in terms of unstructured data and insurance?  

SAI: Yeah, so I think the two-part question, your first question is, what do I need to do, the first thing is, every carrier has their own structure, their own format. So, your first step is rationalizing into a common format. I think the brokers are suffering because of that. I have a customer here, I have five different loss runs. For me to really get a complete picture, that is my focus, it is once you extracted, how do you rationalized to a common format a policy type in this one and this one? So then only you can do inside. So a lot of companies, if you do not do the rationalization, I give you data in a different format, which is also going to die. I think if insurance companies start doing their APIs, you already have structured data that you are converting into a PDF and giving it. Now, if you start giving the API directly, but still my value is the normalization. So that is the first part.  

The second part is, am I going to start thinking about images and video? See right now, again, it is a use case to use case, if you are going to look at a use case, for example, coverage gaps, I was thinking about, okay. Somebody, I can go and take pictures and upload, then I can immediately say, okay, you have some valuable paintings. Based on the images, can you recommend any additional things? There are so many things, but image recognition, like Google and Microsoft are already much ahead of that. Or even the NLP natural language processing. If you are a technology company, you can never beat them, they are much ahead. What my focus is, contextualizing that data to insurance. I think that should be the focus for me. Yeah. Images, okay. What does that image mean to me from an insurance context? My focus is right now extracting and contextualizing and making meaning out of that so that is a value to everybody. You know insured, broker, or an insurance company.  

JAMES: That is awesome. Well, this has been a great discussion. I am excited about the future of machine learning and artificial intelligence. I am excited about the untapped potential of getting into all these documents and really started bringing some normalization. I used to think there was hope for standardization in insurance. I have given up, right? And I am not a defeatist, I am an optimist. Even with COVID and all this other stuff, I think the now is great, I think the future’s better. Just my general rule of thumb, but I think there is literally zero hope, for any semblance of standardization in the insurance industry. I think the only hope we have is for technology to bridge that gap, and there are solutions like yours that are going to do it. So, hang with us for a second. Rob and I have got a couple of news items we wanted to talk about, and I am going to open up by talking about, the InsureTech market in general.  

Now, look, I do not cover a boatload of funding announcements, et cetera, but there has been an incredible amount of activity in the InsureTech space in general, and I think it warrants some discussion. And this is from The Daily Chronicle. InsureTech Market Is Booming Worldwide, GoBear, Insureon, CideObjects, and more. There are some good research projects. Now this will be posted in the show notes so you can get it. Rob, you have been friends with these folks for a while and I have been getting to meet this really fascinating network of folks that exist in InsureTech. There is a lot of funding activity and it does not appear to be slowing down. Low–interest rates are actually pumping even more money into private equity right now, looking for better returns. The public market is so chaotic and so psychotic that people were saying, you know what, we are going to have a better ride over in private equity. We cannot just leave our money sitting and making 0.04%, in the bank or even, theoretically negative rates, which is just mind–blowing. So, it is creating this juiced private equity environment, and you have been tracking this pretty closely. It has been active.  

ROB: And I think it is almost returned to the pre–pandemic levels, already even though we are still in the middle of the pandemic, we just learned, last night, now that we are recording that the president, the first lady now tested positive for COVID-19. So clearly not out of the woods in terms of the pandemic, but you are absolutely right, James, in terms of, it is money looking for a return. We are in a super–low interest rate environment. We know that insurance companies, as well as private equity, need to find a return. That is how insurance companies make money. They have operational gains and losses. But they can always try to be sure that they are going to get some investment game, so part of it is coming from the companies themselves, carriers, reinsurers, and others. Some of course from VCs.  

And then the other part I think, added onto that is, we have seen some successful exits, right? We have seen some of these, IPOs that are coming that have already happened, like Lemonade that are lining up. We have seen others –  However, they are happening, we have had a series of exits now in 2020, that really is proving the value of these InsureTech’s. So I think, it is clearer than ever that this is a really hot space. So, whether that will continue or not, who knows, but it is not a surprise to me that the activity has really picked back up.  

JAMES: Yeah. And I do not expect it to slow down, largely because of, well, what is the old saying, it is the economy stupid! It is all about the economy and it is all about returns. And as long as returns suck, money is going to flow into private equity. The same thing with insurance. When you are going to have a harder market, premiums are going to come up. If their investment returns are in the tank, they got to make it back up on their premium. And that is one of the reasons you are seeing, of course, some movement in that area. Rob, you have got two pretty interesting news articles. What have you got today?  

ROB: Yeah, I actually want to shout out to a group that has been working. And this is a new startup called B Atomic. They have been around for about a year and a half. Seth Zaremba and Sydney Roe work there. Actually, a buddy of mine, Pat West just started working there as well, so these are folks that have been an independent agency space for a long time and saw the need for a new technology solution for agents, so they launching this, their product called Neon, they just launched on Wednesday. Ryan Haley had a big show and I am including in the show notes, a link to the YouTube video of the whole launch. It is an hour–long, but they go by the #IndieTech, and then they say it is agency tech because that is that created by agents for agents, so congrats to them. Something definitely to check out. 

And then, a big headline this week was Allstate announced that they are laying off 3,800 employees, that is about 8% of their total workforce, so we are going to include an article in The Wall Street Journal. But you can find it in insurance journals and elsewhere. So, very interesting times. I think this is not the first time we have been talking about layoffs in the sector, James. That continues even at some of the larger companies. So, the economy, more turnover really, of employees we know that insurance is a very tenured workforce, and we know that, as people leave the industry, whether it voluntarily or involuntarily, then they are not all going to be replaced by people. And AI is certainly going to fill the talent gap to some extent within the insurance industry.  

JAMES: Yeah. It’s a pretty wild ride that auto went through and is still going through because they had this sharp reduction in driving and then this massive return and driving, and then they thought they were going to have a lower total claim dollars, but then severity went up. And so they really got like, I almost feel like they moved too quickly to rebate people. They really did, because they distributed a billion dollars to customers through their shelter in place, payback plan, a billion dollars. That was premature because they had fewer auto accidents for a very short period. It might have been a month and a half. But then severity comes up, then driving comes back in force and they are kind of caught flat–footed. They had handed out all this cash. And this is just two months after they bought National General Holdings Corp for 4 billion in cash. So, they laid out a billion to the customers, 4 billion in a cash acquisition, net income increased 50% to 1.2 billion in the second quarter. It is a weird story; I follow these pretty closely and so do you. And I still cannot quite make heads or tails of all of it. I do not know, Rob.  

ROB: Yeah. Interesting times. We have been talking about automation with Sai so I am kind of curious. Sai you know, AI getting rid of jobs, that is a been a big talk, so I am just curious if you have any thoughts on what the future of AI is going to be? Are we going to be a jobless economy or, what is the impact of employment in the insurance industry from your perspective?  

SAI: Yeah, very interestingly, I do not know whether you read this, Harari‘s Sapiens and Homo Deus book, which is an immensely popular book about what happens to human beings because of all the automation, right? It is one of the fascinating books. He gives an interesting perspective. Do not focus on jobs, focus on humans, human needs. So, it is not about, what are their needs, so this whole book is all about, sometimes I shut down my insurance and I start reading these books to see a completely different perspective, as a human angle. So, automation, and it is just not about one industry. He is asking one fundamental question that looks, planet earth cannot survive, the way it is rapidly growing. So, homo sapiens, the reason why so many species have operated, but we existed, and we are thriving, what is the reason? What is going to, what can it not sustain, so that there is what happens and how this automation is actually a threat to the whole of humanity.  

And if you do not do anything about that, the homo sapiens are going to be gone. And then he is talking about homo Deus, who are the human gods. They will destroy the sapiens because they do not need… What is the purpose of a human being? It is interesting whether it is right or wrong, but there is an interesting view, so to answer your question, I think in the next few years, people are looking for more help rather than replacing a lot of jobs initially. But I think that is a broader question of the effect of automation and the human species, is what is recently, is actually a lot of interest to me. And I am not an expert. I am learning a lot reading those things and it is fascinating. It is fascinating, as a whole concept.  

JAMES: Yeah, I tend to be utopian and not a dystopian, on the future. I tend to believe that things are going to be getting better, that the future probably looks a little more like Wally than Terminator. We will see and only time will tell. Elon Musk issued a lot of warnings around automation and AI. And he is a pretty smart dude. One of my favorites quotes of Elon Musk, the greatest, and you will appreciate this as an engineer, and we are going to close on this. He said the greatest engineering minds of the most recent generation to enter the workforce have been focusing all of their efforts on how to get people to click on an ad. And he said we are focusing the greatest engineering minds of our generation on getting people to click on advertisements when we should be solving big problems. That is why he is trying to get us to Mars. He is trying to solve tunnel boring, he is trying to solve electricity, solar and electricity and we have to focus. I just finished his biography, I thoroughly enjoyed it. And he really challenged me to focus on what matters. Focused all the engineers. We have almost 180 engineers at JBKnowledge. We got to focus our time on things that matter. On that note. It has been an absolute pleasure speaking with you, Sai, best wishes to the business and we are excited to know you and meet you, and thanks for being on the show today.  

SAI: Thank you.  

JAMES: And as always, the most interesting man in insurance, Rob Galbraith. Always good to see you, my friend.  

ROB: Great to see you, James, Sai, great to have you on the show, and yeah guys. Another great episode.  

JAMES: Yeah, rock, and roll. And for all you out there in listener land, this has been the InsurTech Geek podcast powered by JBKnowledge. It is all about technology that is transforming and disrupting the insurance world. I have been your host, James Benham, JamesBenham.com with co-host Rob Galbraith, endofinsurance.com. Big thanks to Jim Greenlee, our Podcast Producer, Kara Dalton-Arro, our Creative Producer, and Adéle Waldeck, our Transcriptionist.

Thank you for joining us today.

We are taking you on a journey through insurance tech, so enjoy the ride and geek out.

See you next time!