Innovation in insurance

Insights on the current state of the insurance industry, and some use-cases for artificial intelligence.

How is the commercial insurance industry reacting to the seismic shifts caused by advances in machine learning, artificial intelligence, and big data?

In this podcast, Frank Wammes discusses these questions and use-cases of artificial intelligence with Ron Glozman, Founder and CEO of Chisel AI, and Gunjan Aggarwal, AI and Analytics Director, Capgemini North America.

Also available on:

Transcript:

Frank Wammes:   

Welcome to a new episode of the Applied Innovation Exchange podcast brought to you by Capgemini’s Applied Innovation Exchange. Insurance is seen as one of the most complex professions in the world with providers having to analyze multiple data sets and variables to offer the insurance product that best suits each customer’s needs. Claims processing is equal or maybe even more complex. We are seeing rapid advances in artificial intelligence and machine learning and how those technologies are causing a disruptive shift across the board for all sectors, from retail to automotive to manufacturing and beyond. So how is the commercial insurance industry reacting to this seismic shifts? Are insurance providers being proactive or reactive?

Today we are looking at how emerging trends in artificial intelligence are reshaping the insurance value chain. And on our panel today, we’ve got some amazing guests. We’ve got Ron Glozman, founder and CEO of Chisel AI, which is a Toronto-based AI solution provider for the global insurance industry. And Ron gave me already some insights on the great prizes they actually have won in different part of competitions. And next to that we also have Gunjan Aggarwal, Senior Director of analytics and AI at Capgemini. My name is Frank Wammes, I’m the Chief Technology and Innovation Officer for Capgemini, Europe. Ron, Gunjan, so glad to have you on this podcast. Welcome.

Ron Glozman:       

Thank you. Excited to be here.

Gunjan:      

Thanks Frank.

Frank Wammes:   

Yeah, super. And what a day. Actually, I think I looked outside today and it’s the first day that actually it’s spring and perhaps spring is also coming towards artificial intelligence when it comes to the insurance industry. So let’s see if that really is the truth and that’s the thing I would like to explore with you. So if we look at the insurance industry, like I already said in my intro, it’s actually a very old industry and I recently saw a movie that actually Dutch, that in the Netherlands, the insurance industry actually found this adoption. So it’s a long industry, it’s an old industry. And we have found that in the past, traditionally it was quite slow in adopting and adapting to technology changes. Ron, what is your perspective? What do you think of the current state of the commercial insurance industry? And perhaps can you give a little bit of an overview what Chisel’s role is in this play?

Ron Glozman:       

Yeah, so I think it’s actually an amazing time to be in the insurance industry. It’s ripe for innovation. Many people will tell you that insurance is one of the oldest industries around, 300 plus years old. And it hasn’t really changed in the way that the business has done. It’s still very manual and still very paper-based. And in fact it’s also, some would argue, unaffordable. If you look at the statistics, less than half the people on planet Earth have insurance. And so one of our missions here at Chisel is to enable more people to get the coverage and the peace of mind that insurance offers and to really make it accessible.

And one of the ways we can do that is by lowering the barrier to entry. And part of the barrier to entry is the manual processes through which a lot of the insurance brokers and carriers today are servicing their customers. And so some of the opportunities we see in the market are better ways of delivering smarter underwriting, knowing your customer, asking them less questions. And so that’s where Chisel AI focuses. We enable insurance companies to automate and sometimes simply speed up and not fully automate some of the routine manual processes such as policy checking, underwriting and claims processing.

Frank Wammes:   

But it’s a very interesting perspective that you give in. And actually I was quite shocked by the amount of people that actually are not insured because we always think that we are over-insured. And of course of the cost. Is it then that you look at, because there are countries where perhaps there is no financial stability that the big new market is or is it also because of actually the cost of setting up and managing the insurance companies in well-established countries where still big opportunities are.

Ron Glozman:

It’s a good question and you hit the nail on the head. Mostly it’s due to the economic stability or the economic development of the countries rather than the cost. But I will say that we’re starting to see, especially in commercial lines that a lot of the companies are insured, but on personal lines, that’s not true. And that’s where some of these new innovations are coming forward is when you’re only collecting a small premium let’s say under a thousand or under $5,000, when you’re paying somebody several tens of dollars an hour, the amount of time that you can spend actually servicing their clients is very, very small if you want to make a margin. And so our value proposition, and I think not just ours, but in the ecosystem, the value proposition should be to enable insurance companies to deliver a better service in the same amount of time or less than today.

And what that will end up doing is it will allow them to first of all increase their margins but also potentially decrease the actual amount of premium that is being charged because they’re not spending as much time focusing per client, but actually delivering a better service. And so I think the opportunity is going to be both in enabling other countries to become more economically stable but also reducing the cost of the actual paperwork to get that insurance.

Frank Wammes:   

Yeah, and more margin. And you know, the question… Eric [inaudible] wrote a very interesting piece where he said like, monopoly sometimes are not bad because once we go into really commoditized world like then it becomes a margin play and if there’s no margin there’s no innovation anymore. So getting perhaps indeed a little bit more space within the industry to really do more innovation that probably also will help increase then probably the experience for consumers and the ability to potentially indeed deliver additional services. Is that also something that you encounter? Is that something that you strive for?

Ron Glozman:       

For sure. One of the things that I’ve come across in my time working in the insurance industry with some of the big players is that exactly. Sometimes it’s about being able to customize a product. Today insurance has traditionally been, I have a product offering and you get it within a package and many people don’t have the ability to customize that package because that package is sort of the average of what the average consumer would need and might not be exactly unique to them. So part of this is going to be how do we enable more products to hit the market and more customized products.

We’re in the era today of everything being online and digital. One of the best examples of that is what Uber has done to the taxi industry and how everything is now one click away. We envisioned that insurance is probably going to be there in five to 10 years where you’re going to be able to go online and buy insurance in three to five clicks without having to fill in all of these long complex forms because most of that data is already accessible through third party APIs or other methodologies. So then it’s the insurance company’s job to actually customize their product to better meet your needs rather than offering you something right off the shelf, which is good but they can do better and we want to push them to do better, which will end up being a better client experience as well.

Frank Wammes:   

Yeah, I definitely agree. And I think I call it the segment of one as a major trend. Like you want to be really having the experience that you have something that is uniquely tailored to you. And that is something that I see in more and more around us. Well artificial is also, it’s really is all around us. We see it deployed everywhere or sometimes we don’t even aware that artificial intelligence is somewhere in some of the services or the products that we deploy. Whether it’s in the business, the vehicles, the homes. I always use my Apple watch is a very good example of like all the calculations that can be done because of all the AI which is around it. But how is the insurance industry reacting to those advancements that AI is in perhaps in some other industries or products or whatever that they could take advantage of or also how can we apply it to ourselves? And the question to you Gunjan, could you highlight a couple of real life applications that you see being adopted of AI in the insurance industry?

Gunjan:      

Oh sure. So AI is reshaping the insurance industry. As you mentioned, this particular industry was very traditionally slow to adopt, but now because of the AI we are seeing a lot of changes happening and which is overall enhancing the customer experience and even bringing, generating a lot of revenue for the insurance sector. Now in the highly regulated sector like insurance, what is more important is the AI does not end up in a black box. The AI explainability is playing a very crucial role, especially for the complex non-linear models which are very much part into the insurance sector. And as AI explainability is becoming more and more available, more and more transparency in the system because of that, it’s enabling various use cases.

The use cases, what we are saying is like let’s say computer vision techniques, it has a huge application in the PNC LOB. So take an example, automotive policy holder could just take a photo of their car after a crash to estimate the damage and find the closest repair shop. So that’s a very beautiful use of a computer vision techniques and AI into that PNC. Second, as you mentioned, the more of a personalization, let’s say next best action is every customer is unique. Their needs are different. The needs are at a different stages of the life even. So to enable that next best action is not just what is lying within the organization, even bringing lot of external data sets, let’s say your fitness monitor data, your card IoT data, bringing everything together and enabling the next best action. Even doing that AI like a natural language processing on the unstructured data, maybe email communications or some more is making that whole personalization and cross-sell up-sell more, more and more unique to the customer.

Not only that, but even in the underwriting process as Ron mentioned, lot of for automation is happening so that more risk can be calculated in that process. So there are various use cases which we’re seeing from different LOBs, from the claims distribution, underwriting, pricing in the insurance sector.

Frank Wammes:   

But now is the insurance sector of course also a very sensitive one. I expect like, I want you to support me the moment that I need it. It’s a security almost that we buy for unexpected things. But now you talk about AI and more and more customers also start realizing that AI is all about data and more and more also we have discussions about ethical AI and is AI going to help us or destroy us. To what extent, if we now have so many different applications in this particular industry do you feel that that could potentially have a backlash or a concern from the consumer towards industries that will apply it? Or is it so much in the back engine of the interaction that you haven’t seen that yet? Ron, I don’t know. Perhaps you can give your perspective on that.

Ron Glozman:       

Yeah, I think it’s interesting because one of the things I’ve seen, at least here in Canada and now I can’t speak to the backend being AI because I haven’t personally examined it, but one of the things one of the carriers here in [inaudible] is doing is on personal lines, specifically auto, they now have this new capability where if you install an app on your phone, you can actually reduce your premium if you’re a safe driver. And I think that’s very, very cool. And it’s one of the first implementations I’ve seen in production, at least where a company is leveraging next level technology. Now it might be more an expert system and rules-based rather than a deep learning model, but either way it’s an implementation of a new technology to an old market. Car insurance has been around for many, many years to give you different pricing based on your driving characteristics.

And I think it’s very cool. And in fact we’ve tested it out in my family and two of the family members were able to get a 25% reduction on their premium. And if you think about that, 25% is a lot of money for some people considering that the premium is several thousand dollars. So I think the insurance industry has actually been quite fast to adopt it. In speaking with one of our customers and partners, they gave me this analogy where insurance is still using fax. Almost all services except governments have stopped using fax, but a lot of insurance companies still do it, but they’re also leaping straight towards AI. And in the middle there was the email and portal era, which some companies ended up going into but many of them are completely skipping that and going straight to AI.

And for an industry that is very risk-averse, I find that surprising. But to the points made earlier, I also think that being able to describe it and not have a black box is part of the reason that is happening because they’re able to have some confidence and some understanding in what is happening in the back end rather than just taking a leap of faith. And I think that is the biggest catalyst of why the uptake on AI has been higher than you would probably expect.

Frank Wammes:   

Okay. That’s a very interesting perspective. By the way, I’m really curious now of course, what kind of car you drive and whether you were one of those two or that you are the one that actually is going to pay the premium of course now. Because that of course comes with it, but I agree with you. I think it’s the transparency probably behind it because it’s clear that if you don’t opt in into this app then potentially you are worse driver. The question is do we want this or is it much more like these kind of economic decisions could potentially also then have an influence on the behavior of people? Are we going to shape society perhaps because people are going to be more aware of like, hey, I can save money now rather than in the past it was just one option and you just better hope you didn’t had an accident and did we change costs.

So there’s a little bit of a funny element to it that potentially AI could change society much more because people will change their behavior. Not so much because of the AI, but much more because of the transparency that comes from AI, which is a little bit in contrast with that we always think it’s in-transparent. So that’s interesting, Ron. You also touched a little bit now of course with this app. That means that basically you have connectivity to it and we see this explosion of data coming from connected devices. So how do you see that insurance providers can really use the data from an IoT perspective to improve customer satisfaction? Do you have both some examples? Gunjan, I don’t know, have you have some perspectives on this IoT with AI and insurance providers?

Gunjan:      

Oh yes, Frank. So there are many use cases of IoT in the insurance sector. So let’s say you have home sensors which can monitor fire, wind, any water damage, which is very useful for the insurance industry. As Ron mentioned, vehicle sensors like you can save a lot of money on that. How’s your driving pattern? Isn’t it dangerous driving pattern? Based on that, your insurance can be calculated. A fitness monitor, we all wear these Fitbits, Apple watches. They all track our health activities. And it comes for a more like a win-win situation. The customers will get some benefits out of it if they are tracking actually their fitness. And even the insurance industry, it helps a lot by merging all this IoT data with the other data sets which are available and that are more holistic view of the customer.

Frank Wammes:   

Excellent. Ron, do you also have some examples or do you have different perspectives on the Internet of things and the insurance?

Ron Glozman:       

Yes, thank you Frank. So unfortunately I can only speak from secondhand experience since we’re more on the software side of the business rather than the hardware, our IoT side. But in my travels around the world and in competition at many of these startup competitions, I’ve come across many companies offering IoT devices. And I think I’ll follow up on the previous point regarding how sensors and flooding. So I know that one of our partner companies that we work with, about 49% of all the claims that they receive in their property business is due to flooding or water damage. So 50% of all claims. And so any type of sensor that you can install to detect a leak, to detect changes in room humidity, to detect moisture through the air or a pipe is going to do wonders both to reduce the actual costs and hopefully prevent the claim.

And there’s one company that I know of in the market who says our view of success as an underwriter, so they’re actually a startup who is insuring houses, is reducing the amount of claims that come in. If you can buy an insurance property and we can help deliver you an IoT device that will notify you of a leak or some other type of loss mitigation ahead of time, it’s a win for the customer because their premiums are going to stay low because they’re going to have no claims in their history. And on the carrier side, it’s a win because of course, 70 to 80% of a typical premium dollar will go to cover claims. And if we can bring down that loss ratio, it’s a win, win, win on all sides. And so I personally am very excited for IoT devices.

I think the one thing that is going to be really, really key is privacy. And so that’s where I think the European and lucky you, you live in in Europe, you get to see the GDPR in effect. Unfortunately in North America, our data privacy laws are not quite as stringent or user friendly. But I’m hopeful that we will follow suit and also implement the right to own your data because I think that’s key. IoT devices are great but you need to also own your data. And so that’s more of a legal issue and a political issue that insurance companies are going to have to be conscious of.

Frank Wammes:   

I totally agree. And the question I had in the end is it not going to be a market for us as well because people will move towards companies that actually comply perhaps from themselves already to certain of these kinds of aspects. So I totally agree. But one of the other things of course, which is very crucial is if you are, and Ron, I think you gave a good example that people sometimes immediately jumped from facts to AI. But you then also need to have the skills. So you need to have a workforce that is assisted by perhaps rather than replaced by technology. But you have to be able to deal with it. So what kind of skillsets do you both see are required in this change within the industry? Gunjan, do you have a perspective on this?

Gunjan:      

Yes, Frank. So as I said, that more and more explainable AI is coming up. It’s actually the workforce has to be assisted by the AI. So when I say that it means, let’s say AI has come up with a result which is required for regulators. Now in human terms, why did the machine has come up with this answer? Okay. That’s where that innovation, which is happening in explainable AI, like sharp values. That’s moving from that traditional features which gives little or no insight to the workforce, it’s actually helping the workforce to understand what is the reason with this particular algorithm output. And this sharp values like take all the samples, do the feature extraction and easily helping the regulators to understand that what is happening behind that algorithm. So that’s where is not actually the workforce is getting replaced by the technology. It’s more of assisted by the technology because of all these explainable AI innovations.

Frank Wammes:   

But it means that the workforce there needs to be open, needs to be able to deal with this different way of how the technology is supporting you rather than you just type in the stuff and things come out.

Gunjan:      

Yes. There is a definitely a shift. The workforce as you said, like it’s more about working closely with the AI, is not like replacing with the AI.

Frank Wammes:   

So if then companies, they take this decision, they are going to build on the different capabilities that they have. We have the different kinds of use cases. What are the core elements in defining a successful AI strategy then? What are the things that company really need to do? And Gunjan, I would like to start with you because then I would like to end with Ron, because of course he has now a very successful AI company. So I just want to hear your perspective Gunjan and then Ron like you really need to give the answer. So Gunjan, what’s your perspective?

Gunjan:      

Yeah, I’ll keep it very short and crisp. I would say the first is have a clear vision, clear use cases, what needs to be done powered by the right talent and the right technology infrastructure. These are the four elements which are very much required for any successful AI strategy.

Frank Wammes:   

Good. Ron, you have been successful. So now just give us the magic.

Ron Glozman:       

Well, so I’m happy to answer that. I’d just like to add my thoughts if you don’t mind to the previous question because I think that’s the right view. And a lot of times people are fearful of AI because they think they’re going to be replaced. But I can tell you from my experience and our customer’s experience. So when I go to their office and I interview people, and I’ve done this, and this is one of my favorite sort of activities that I get to do in my role as CEO, is solicit feedback and really sit down with the users, understand what they like and what they don’t like. And here’s what I’ve come to learn.

Often times, let’s say we’re talking about an underwriter, they are working not 9:00 to 5:00 but sometimes 9:00 to 7:00, 9:00 to 8:00 or sometimes the opposite, 7:00 AM to 7:00 PM because there is simply more to do than they have time to do in a day, which means that those two or three hours that they’re working overtime, they’re missing their daughter’s first ballet recital, their son’s first baseball game. They might be missing an anniversary dinner or let’s say the love sports, they might be missing out on watching their favorite football team. And so our mission at Chisel has been for the longest time to help people work smart and enrich their lives, which means if we can assist you in doing your job with AI so that you get to go home at 5:00 PM that is a win because we’re giving your precious family time back. And so it’s 100% not about removing people. It’s about enabling them to work smarter and more efficiently and get the work done in a reasonable amount of time so that they can go home.

Frank Wammes:   

Now I understand why your company is called Chisel because you just give me the chisels Ron, in a very easy to use manner. But I agree, and I have never had somebody articulate it this well. So thanks for that. I think indeed, this is an opportunity not so much a threat. And when it comes to the AI strategy, so what do you think next to explaining the opportunity, what are some other elements that you think companies need to have for a successful AI strategy?

Ron Glozman:       

Yeah. So one of the things that I’ve seen that works relatively well as an innovation team or department or the opposite, which is empowering everybody to be running projects and proofs of concepts, which is a good first step. But a lot of companies fall into fallacy where they have an innovation department and they want to do a pilot, but they haven’t changed their procurement process. And as you probably know, these insurance companies have very, very strict procurement cycles very, very long. And for a company like them, 12 months might not seem like a long time, but for a small startup slightly earlier than we are that is bootstrapping and doesn’t have venture funding, they cannot afford to go through a full procurement cycle.

One of the things that companies need to have as part of their strategy is they need to have a different procurement cycle for pilots and proofs of concepts. And that cycle should ideally be like one month. And that’s really where you’re going to find success because a lot of the small startups will not be able to support the long sales cycles of procurement cycles that need to go through.

And then I think the second piece that I’ve seen work very, very well is having a budget. Of course doing proofs of concepts and pilots should not be cost prohibitive. So I’ve seen companies that especially startups that asked for very, very large proof of concepts and pilots. And in my opinion that’s not reasonable. But on the flip side, the companies, specifically the brokers or the carriers or the reinsurers that are participating should have a stake in the game, which means having some budget dedicated aside to this. Oftentimes I think it’s improved over the past two years. We would come in and they would need to find the budget.

So they want to work with us. They see that there’s a business need but they don’t have a budget and they need to go out and solicit it from departments, which again can take months and months of time. So if I can make two suggestions on how to improve corporate strategy for AI, one to reduce the procurement cycle and actually have a separate stream with much shorter compliance questionnaires, etc. And the second one is to have a dedicated budget set aside. It doesn’t need to be huge, but it needs to be there and it needs to be at the discretion of the budget holder so that they don’t have to get approvals to deploy the budget unless it’s above some threshold.

Frank Wammes:   

Excellent. Yeah, I completely agree. And one of the things that we always said is like your whole organization needs to be digital aware. And that not only goes for some people in the IT part or from the commercial part, but it’s actually indeed like your support staff, risk, procurement, finance because you’re going to craft new business models, you’re going to procure in a new way and you need to do it in a different mode. Excellent. You know, I learned so much again. I’m a user for the insurance industry. I’m not so much working with the insurance industry, but for me it is clear that this is one of the industries that has a tremendous potential because of perhaps sometimes the lack of investments in the past or advancements on technology in the past, but now taking leaps forward, a good example, Ron, that you gave from going from facts to AI in one go.

I think the use cases are very valid, are very rich and they come in a lot of numbers already and it also learned me, and that was a little bit of an insight that came out of it is also probably shaping our culture and the way that we think because of the transparency. And that’s the fun part because indeed, AI doesn’t always have to be intrasparent but actually becomes more transparent on how we spend things, etc. But it is about the opportunity and not about the threat. It’s about creating more value and this notion and need of being able to see your first ballet recital from your kid. That made impact, Ron. And it’s indeed about how do you test it, how do you test it in short cycles and that also needs to be supported of course because you’re going to probably do it with different people.

I would like to thank you. I think it was a fascinating discussion with a lot of rich insights that people can build on and think about what it means for implementation at clients. And I hope everybody enjoyed this episode of the Capgemini Light Innovation podcast, and you can connect with Ron. Ron, where can people find you on social media?

Ron Glozman:

They can find me on Twitter @RonGlozman, G-L-O-Z-M-A-N, or @ChiselAI, and connect with me on Linkedin. If anybody wants to reach out, I’m more than happy to add you to my network and answer any questions.

Frank Wammes:   

Well, excellent. And I think you can deliver a lot of value to the people. Gunjan, where can people find you?

Gunjan:      

I’m on Linkedin with the name Gunjan Aggarwal, and on Twitter it’s gunjan_amit.

Frank Wammes:   

Excellent. Well, thank you so much for your participation and the richness that you brought to the discussion. My name is Frank Wammes. You can find me on @fwammes on Twitter, and fwammes is on Linkedin, and I hope you will join us next time again on the Applied Innovation podcast, brought to you by Capgemini’s Applied Innovation Exchange. Thank you.