EP 65 - How Can Trust and Tech Work Together to Redefine the Insurance Journey? - Manjit Rana - EVP Insurance at Clearspeed

Speaker A: Hi, this is Michael Waitz. And welcome back to Insurtech Amplified. Fraud prevention and customer trust are no longer separate conversations in insurance. They are now deeply intertwined. And as fraud grows more sophisticated, which we will get into, insurers must find ways to protect themselves and their customers without adding friction or eroding that trust. And at the same time, customer expectations around speed, transparency and fairness of across the value chain have never been higher. In this episode, we hope to answer some of the industry's most pressing questions. How has the face of insurance fraud changed, particularly in the post pandemic world? Why is traditional approach to fraud detection no longer enough? What innovations, whether from existing insurers or from insurtechs, are enabling the industry to proactively detect fraud while keeping honest customers at the center of their insurance journey? To explore these questions and more, we are joined by Manjit Rana, uh, EVP of insurance at ClearSpeed, who will share his perspective on how trust, speed and simplicity can reshape fraud prevention and customer engagement across the insurance value chain. Majin, thank you so much for coming to the show. How are you doing today?

Speaker B: I'm great, thank you Michael, and a pleasure to be here. Thank you for inviting me.

Speaker A: The pleasure is all mine. Okay, let's jump right in. Why is trust now the critical currency in the insurance space?

Speaker B: I think it's not just in insurance. I think that the whole aspect of us trusting organizations, we work with, um, you know, product providers, we buy services from the trust is really important. Um, if somebody, you know, if you're watching and you know, 100 meter rates, you need to feel confident that you can trust that person's genuinely not taking performance enhancing drugs, you know, and they're being fair with, with everybody else that's in that race equally. You know, if, um, you're on a dating site, for instance, I'm not condoning it, but if you're on a dating site, for instance, you want to know, you can trust the profile. I said that I did have a dating business previously. Um, but you can trust, trust the profile that you're, that you're looking at. If, if somebody's asking you to, you know, romance fraud, they're asking you to transfer some money because they desperately need a bit short term cash or something, have to be able to trust the person that, that you're having the conversation with. Insurance in particular though, as a consumer, I need to trust my insurer that I'm paying the money year in, year out and at the point where I have an issue where I need the insurance company to step in. I need to be able to trust that they're going to be there for me and they're going to help solve, resolve the problem. Equally as an insurer, um, I need to trust the fact that the information you're giving me is genuine and fair, um, and accurately represents the risk that I'm effectively underwriting. Because if you're not telling me the right information or you're giving me false information, I'm pricing the risk incorrectly, um, which impacts everybody else that I also sell insurance to. Um, and equally when you make a claim, I've got to be able to trust the fact that what you're telling me is accurate. You're not trying to defraud me. You have an exaggerated information. So I think the whole trust thing is, is there's lots and lots of reasons why it's becoming almost like the new currency, um, in, in the commercial world, um, regardless of what industry you're in.

Speaker A: Sorry, I want to follow up with this if you don't mind, but I want to do it in two pieces.

Speaker B: Sure.

Speaker A: I'll get to the insurance thing. Second, if you don't mind. But why do you think from your perspective that the trust has been eroding over time? Because I feel the same thing. I feel like back in the 1980s and the 1990s I wasn't so worried all the time about the information that I was getting. And now today I look at everything and I'm just like, why has this happened, do you think, at scale, not just in the insurance industry, but just kind of across life in general?

Speaker B: I think there's probably a whole stack of reasons. Um, the way technology's evolved, it's enabled us uh, to be able to do things that we just couldn't possibly have done 10, 20 years ago. So um, if I want to create an image of uh, a iPhone 16 Pro with screen damage, I just go into ChatGPT and I ask it to create me one within literally less than 10 seconds it's there me to being able to do that. Ten years ago I would have had to be super skillful graphic designer, a ah, ton of expensive equipment. So I couldn't really do that. So, and, and everybody knew it's really difficult to do. So it, so you, you, you couldn't really do that, do those things. Um, and I think the, the other, the other thing is we as a, as a population we're constantly connected to technology. You know, we walk around with these things glued to our hands so we have access to technology now that we just wouldn't have had access to a number of years ago. And the fact that this, that, you know, if you especially know with the new AI technology, generative AI technology that's around, it's super, uh, easy to be able to do. I don't have to be a skilled person to be able to do it. I literally can go type in something into one of the AI tools and it magically just does the stuff that I need it to do.

Speaker A: So I'm going to say something that I think is going to be very non controversial.

Speaker B: Okay.

Speaker A: I think, I think 100% of the people on this call, I think 100% of the People on this podcast, whether it was 1970 or 2025 or 1990 or 2025, wouldn't have gone to Adobe Illustrator or wouldn't have gone to um, chatgpt and would not have made a fake cracked iPhone to get a claim on an iPhone. And I don't think that has anything to do with our income levels. We just would never do it. So I'm curious, and I'm really thinking about this a lot, like, I'm just curious, like, has the number of people that want to create fraud changed over time? Or is it just the fact that technology has made it so easy to do this that the people that had the propensity to do it now be like, simple now? So I'm going to do more of it. Do you know what I mean? Because you would never do this. Neither would I.

Speaker B: No, but, but that, that's down to our personalities, our characters that are upbringing, whatever. Yeah, yeah, but, but also I think people, the people that commit fraud would still have committed fraud. They would have just done it differently. So for argument's sake, the type of fraud that we might have found previously, um, students and end of it, end of a university year, you know, and a lot of cases where they kind of push the, Buy a, um, a tester pot of paint, 25 pence from a, from a DIY store, spill it on the carpet, make a claim for replacement carpet, and then once, once they've made the claim, they push the sofa back over that, that patch where the paint was. That's still fraud. It's not using technology. But they, they know the rule, they know the system enough to know what they can get away with. I guess the difference today is I don't have to physically damage anything. I can literally use a bit of technology to just do it really quickly. So it's more easy for people to be able to do. Also I think when, when you find the financial climate gets really tight, people are quite tempted to do stuff they would not normally do and not in the normal circumstances. And an example of that.

Speaker A: That's fair. Go ahead.

Speaker B: So, um, the abi, um, association of British Insurers did a report where they, they said that home claims, 87% of them included an element of exaggerated fraud. Now that means I get broken into. They've done, you know, they've taken my patio door through, they've come and stolen my TV and and so I've got physical proof that I've been broken into. There's my big gap on the wall where my TV is missing. I've got a police report because I did get broken into. But when the report goes into the insurance company it also includes a couple of iPads which I do own and I do have receipts for M and my wife's diamond necklace which again she does own and we do have receipt for now those two iPads and the diamond necklace didn't actually go. But my view is I've kind of been paying insurance for X number of years. Here's my opportunity and I've not got anything in return for it. So I think that's a critical thing. I've been paying for something and actually I've never received any benefit for it. But here's my opportunity to kind of even up uh, the situation. So I exaggerate that fraud. Now there's nothing to do with me creating any false evidence or anything like that. It's just really easy to be able to do the challenges as an industry, as an insurance industry we've never really been able to identify that because a, the financial climate hasn't been that bad so people didn't really do exaggerated um, as much as they're doing now. Um, so that market, that market has changed and again typically what's happening now is people designed solutions to identify fraud based on what the situation used to be in the kind of fraud that people used to commit 10 odd years ago. Fraud has evolved as a result of technology evolving, but as an industry we're applying the identification methods that we were using years ago and I'm not really sure they're appropriate.

Speaker A: So let's get into this because that's actually really interesting part of this, right? How has the role of technology changed and then how has that changed the fraud but also how has technology been used to rebuild that trust? You know, because it goes both ways.

Speaker B: Absolutely right.

Speaker A: If I buy a water gun and walk down the street with it, and you're my neighbor, you're going to be like, I'm going to buy a water gun too, and maybe a bigger one kind of thing. So, like, what are both sides of that coin, if you don't mind?

Speaker B: Yeah, so typically, um, the way fraud, fraud used to be detected, predominantly, uh, two ways. So first thing is you use all of the data. You have to be able to profile, using the data, you profile individuals to identify the kind of individual that's more likely to do something that exaggerated claim or potentially commit fraud. Now, you can do that because your data shows, for argument's sake, you're in a certain part of the country, certain income, certain credit score, certain profile. So therefore the insurance company would say, okay, we can't screen everybody. We don't have the resources to be able to do that. So let's pick the most likely segments. Um, can I ask you this, if you don't mind?

Speaker A: Yeah, sure, but hold the thought because I don't want you to lose this train of thought. But I'm just thinking, like, in some cases, is it so blatantly obvious, like a flashing red light, like, this is clearly, you know what I mean, where you're standing around with your colleagues and you're like, all five of you are just thinking, this is definitely fraudulent. Do you know what I mean?

Speaker B: I think a lot of that comes from experience claims handlers, experience fraud handlers, that just because they've been doing this for 10 years, 20 years, they know, uh, this just doesn't feel right. Now, to be fair, I'm not really sure technology is ever going to replace that. We can augment it, but that experience is super valuable. And the thing is, we have to make sure we retain that. But that's generally. So that stuff gets found because, as you say, it's just really obvious. Stuff that doesn't get found is the stuff that is not that obvious. But in order to find that, we kind of almost have to treat everybody as if they're a potential fraudster, which is terrible. Think about it. You know, you've paid X hundreds of pounds to ensure your home with me, and the one minute where you kind of need me to now return your part, you know, the. My side of the bargain. Effectively, you're, you're essentially telling me that you have to go through a whole load of processes where clearly you don't really trust me, right? That that breaks the relationship down. And if you, especially in insurance, um, because we, we too really, really, we provide an amazing service to people when when we get it right. But if you talk to most people about insurance, firstly that, oh yeah, my friend had a claim and the insurance company never paid out, um, or, oh, they're not going to, they're not going to pay me the full thing anyway. So I've decided to just up, up, uh, and having exaggerated it a little bit just so that I kind of get back to affecting. So there's clearly an element of mistrust. No. And the, you know, if the insurer is saying, I can't really trust you, so therefore I have to almost look at you as if you're potentially committing a fraud, committing a fraudulent claim, um, and the consumer doesn't really believe you're going to pay out fairly, there's a massive breakdown in, in that relationship. I think some of that is fundamentally also to do with the way that we design insurance as a service. So insurance is a safety net product. Yeah, you pay me money, I put a safety net in place. If you don't fall off the tightrope, nothing happens. You don't need a safety net. And actually I don't really want you to fall off the tightrope because I don't want to talk to you because, ah, the minute you contact me after buying your policy, that is going to cost me as an insurer money because something's gone wrong typically. And for the customer. I used this analogy before. I was doing, um, a presentation at a university to a number of students and I asked the question, hey, what's insurance? So I got kind of like the box standard answers. And then one guy stood up and he went, it's kind of like me going into an electronic store and buying a, uh, tv. I take it home, I take it out of the box. And as I take it out of the box, there's a great big sticker on the screen that says, do not switch on for emergencies only. Well, I've just paid £500 for it. So I stick it on my wall, but there's no emergency, so I don't turn it on. And then 12 months later I get a phone call from the, from the store saying, thanks very much, we're coming to collect the tv. Um, and by the way, you can have another one. But it's gone up to 525 pounds. And he says, that's insurance now, the problem. And I can understand the sentiment because he's kind of going, I pay for something, I don't use it. It's not exciting. I don't queue up for two hours to try and buy it. It's not exciting. Um, and if I don't use it, what did I pay for? And people don't really want to get into the. The reason you buy insurance is if something goes wrong because most people don't believe their home is going to get broken into. Most people don't believe they're going to have an accident. Most people don't believe their car is going to get stolen. So how do we add value? And I think this is. This is probably a much broader, longer conversation. But if we change from the safety net proposition to a guardian angel proposition as an example.

Speaker A: So what's the difference?

Speaker B: I'm, um, driving down the road, I'm coming up to a bend, and like, literally five seconds before I, you know, a voice pops up out the. Out of my car saying, manchu, just take your foot off the accelerator just a little bit. Now that, because it's happening at that moment in time, I'm probably going to respond, and you've probably prevented me from having an accident, or I may still have an accident, but it's not as severe as what I would have had if I hadn't taken my photo, et cetera. Now that is because if. If my insurer was doing that for me, I'm getting value multiple times in a day. So my relationship with my insurer is going to be very different.

Speaker A: But this is changing, right?

Speaker B: I have to trust my insurer to deliver that. Right.

Speaker A: So I want to talk about this because I do think that the technology is changing this. And I want your opinion as well, just based on stories that I've heard and actually, let me reflect something to you before we get to that.

Speaker B: Sure.

Speaker A: I had an operation. This is now a few years ago, and I didn't have any health insurance. And the reason why was because I had worked at Morgan Stanley and Goldman Sachs my whole life, and I didn't even know how to buy. Buy insurance.

Speaker B: Right.

Speaker A: I did. I wouldn't have known what to buy because it was just provided for me. And the insurance that I did have was so comprehensive that I never even had to think about it. Like, I never had to think about it. But once I had this operation and it was expensive, I was like, okay, this is stupid. You need insurance. And I bought it anyway. And then every year in January, I have to pay my premium. And I actually get nervous because I feel naked without my health insurance.

Speaker B: Right.

Speaker A: I've not used it since, and I don't care. And I don't think this is an income thing. Sure. M. I'M not using it, but I'm super happy that I'm healthy. But I'm also happy that if I fall down or if I need some help, I'll be able to get it. So that's. First of all, second of all, this idea of insurers being like a guardian angel, I think you can see this kind of across the insurance industry, right? Like, it's not just in health insurance. It could be in your car insurance, it could be your. Somebody was telling me during COVID that their insurance company called them and I think they had a telematic device installed in their car and said, you haven't driven your car in like three months and your battery is going to die if you don't fix. If you don't do it, why don't you go turn your car on, at least take it for a little spin and then bring it back so the battery doesn't die. And while that in and of itself doesn't seem like a huge service, first of all, it's more than the insurance company was doing before. And second of all, now that guy is going to start falling in love with his car insurance company because he's like, uh, just the hassle of having to get a new battery. It's not even really that expensive, but the hassle and the time, like, that's an awesome service. And they could do the same thing in the health space. Sorry, go ahead.

Speaker B: But that relationship that they're building with that consumer at that moment in time, because the alternative is the consumer's looking out of his window on his driveway and kind of going, that damn car has been sat on the drive for three months. It's not moved. What am I paying my insurance for? Um, like the insurance going to give me a refund for the time that I've not been using it. To be fair to some insurers, they, they did give. Proactively, did give refunds. Yeah.

Speaker A: I mean, there is UBI as well. Right. So it's a different metric, but yeah,

Speaker B: so, but so I think it is about that. Ah. If I, if I trust my insurer, am I less likely to defraud them? And equally, you know, as an insurer, if I'm trusting the consumer, you know, when they change something, um, you know, put upgraded M wheels on the car or something, they're going to ring up and go, hey, Mr. Insurer, by the way, just to let you know, I've done this. Um, so that trust has, you know, as we said, that trust has to work too.

Speaker A: So what are the other places along the policyholder journey. Right. Where you can have these little instances? Uh, because look, one of the guys in the fintech space said to me, a bank, you interact with your bank every single day, whether you're taking money out, putting money in, using it to pay for things. You use your credit card, which is generally associated with your bank. So you have a relationship with your bank whether you like it or not. But in general, there aren't a lot of touch points for insurance companies. I only talk to my insurance agent in January. She's great, but I don't need her in March, really, or in June, as long as nothing happens. So where is it? Because the trust is really important and I trusted her, which is why I bought insurance from her, because she's a friend of a friend or whatever.

Speaker B: Right, right.

Speaker A: And I'm pretty confident that if I need to make a claim, I'm going to be okay. Although I don't know that. But where, in your experience along this journey, where are the other places you can create these touch points? Not every day, because that could be annoying. But enough of time where I'm like, yeah, I trust that guy or I trust that gal, or I trust that company because I've just been interacting with them all the time.

Speaker B: And um, so I think if you look at, um, look at household as an example. So if I leave my house and I've left the iron on, I get an alert automatically from my home security company that says, you appear to have left the house. You do realize that the iron is still on. Um, now I don't have a problem with the fact that someone's monitoring and looking out for me. Would I be happy for my insurance company to do that? I'm not so sure because the minute I window or door, door open, um, and they let me know, I'm kind of thinking, oh, hang on a second. This is kind of like an ideal opportunity for my insurance company not to pay the claim because I did something wrong. It's, it's very similar to telematics insurance where I'm being tracked in the, in the car now.

Speaker A: Yeah. Because if you're speeding and you get into an accident like you were speeding.

Speaker B: Yeah. My, my car knows when I'm tired. It tells me I. My car knows if I've changed lanes because it tells without indicating. It tells me my car tells me if I'm driving too fast. I don't have an issue with my car and the car manufacturer knowing that. Would I necessarily want my insurance company to know that right now, probably not. Um, so there are lots of opportunities. I think there's this relationship that creates that friction between my insurance company providing that service and somebody else providing that service. But there are a number of touch points where I would get value from that on a daily basis.

Speaker A: Let me ask you two more questions about this. I have more, but just two things I want to put them together. One is, do you think most people who buy insurance know that that claim. Not the claim, sorry, that the premium that they pay is not just like taken putting into a little box that says like, this is Michael Waits premium and if something bad happens, we pay his claims out of this premium. But that they actually take that money pool together and run some of the largest and most sophisticated investment funds in the world and that they're investing in alternative assets. Some of them might even own crypto card like that. They're just running massive investment companies and that they're also using some of the returns from there to say, oh, well, we made this amount. We had an 8% or 9% return on average over the last 50 years for our investments. And when a claim comes in, we're not just like taking money that you gave me out of a box and paying a claim. We have a massive balance sheet that we're running. First of all, do you think people know that? Most people.

Speaker B: Honestly, no. I think people, generally people's knowledge of the way insurance works is way more limited. And again, I think a lot of it is that if you look at banking, so from a younger age, you will be introduced to banking because your

Speaker A: savings accounts and stuff like that.

Speaker B: Yeah, exactly. Um, so it's kind of given, you know, we grow up with, with understanding how banking, not investment banking, but understanding how a current account.

Speaker A: Sure.

Speaker B: How deposit. Sure, sure, sure works, um, etc. Um, if you ask most people, even when they're at university, do you know how insurance works? Apart from students that are studying ag, you know, um, actual legal studies?

Speaker A: Sure, sure, sure.

Speaker B: Like most of them really, they only understand it at a very, very superficial level. Level. Um, and they just, they see it as. It's something I put money into now. You know, I've had conversations with insurers in the past and said, you know, when you send out a renewal notice, why don't you just include a pie chart that says you're going to give me £700 of, uh, that £700. Do you realize it costs this much to run a business? This much to pay out in claims, this, this much to, to um, compensate for fraud etc. Because then I, I would understand actually the £700 is actually not really that expensive in return for the, the value that I'm getting. The problem is I'm, I'm getting a figure uh, that I'm blind to. And also because it's, especially in the uk because of price comparison size, I will see a figure, uh, anything from £400 to potentially £4,000. So it's kind of like me buying a tin of baked beans. Right. If I think the Branson brand is cheaper today still baked beans, they might taste a little bit different. But actually what is the difference? An insurance product is an insurance product. So we don't really educate the market that well in terms of what's important to, to look for. The UK particularly is super price competitive. I, I heard a figure that we will move from one insurer to another for the sake of £5. I mean in the US I think that's at least a couple of hundred dollars.

Speaker A: God, it's not worth the hassle for me.

Speaker B: I mean that's crazy.

Speaker A: Not worth the hassle.

Speaker B: Move that readily. But clearly it shows that I don't really understand the value between the different products.

Speaker A: I want to get back to the property insurance in a second, um, right now because I think there are some interesting things that could happen. Let's just say I pay for property insurance for my house or for my flat. Yeah. And I want to have some security. Right. So I want the insurance company to provide some service to me. But you're like, I don't want necessarily all that data about my life because let's say I'm dating two people at the same time. I don't want them to see this person come into my place and that person come into my place. That's my private business. But what they could do is they could actually subcontract that out to a company that just does that. So the insurance company never gets that information. You could have maybe an insurtech that just provides that service to a big insurance company. But then the insurance company gives you that for free. So they'll install like a camera, let's just say in a, um, particular place to give you the utmost protection. But they'll never see the data unless somebody breaks into your house and you give the other company permission to do it. But they pay for that. That would be an incredible. Because now no one's ever going to break into your house or it's an edge condition if they do. But now you feel like you're Getting a service as well. Right. So now I'm paying my premium. Every year. I'm paying a little bit for that camera and that other service, if you know what I mean.

Speaker B: Yeah. I mean, if you, if you kind of flip that slightly in terms of. My home security company has all of that data already. Because they're monitoring my home.

Speaker A: They do, but I'm. But I'm paying for it. I think the insurance company could provide the service by paying for that. Yeah.

Speaker B: Okay. So, um, I'm kind of flipping it over and saying, what if my home security company says Manchu. Instead of paying 40 pounds a month for just your home security, we're going to up that to £60amonth and we're going to bundle insurance in underneath. You don't need to know the insurance.

Speaker A: Sure, sure, sure.

Speaker B: So would I feel comfortable with that? Yeah. And actually.

Speaker A: Yeah, because they're not underwriting it, so an insurer is going to underwrite it, but they're actually providing that service to you separately. I get it.

Speaker B: Yeah.

Speaker A: That's interesting.

Speaker B: And the fact that I'm comfortable with them having access to that data. The same thing, you know, if you look at telematics, for instance, you know, Porsche have all of that information about my driving style and everything else. They said to me, you know, as part of your monthly subscription for the car mindship, we're bundling insurance in. I wouldn't have an issue with. I wouldn't have.

Speaker A: You would do that all day. Yeah.

Speaker B: So I think it comes down to who do I trust with that information and why do I think they need it? Now I know my car manufacturers not looking for information, um, to say, does Manjit. Is Manji having an affair or something?

Speaker A: Yeah, yeah.

Speaker B: Not that be brave enough to. But, um. Uh, you know, do I want my insurance company to know that? Your point about. If there was a new insuretech that held that information and then it was made. Made available. I kind of think in Europe and the UK we're less trusting of brands that we don't really know.

Speaker A: That's weird.

Speaker B: Right. Whereas in the US we're way more, you know, way more open to a new brand coming along providing that kind of service. Um, but if, if it was done by, I don't know, in the uk, if that service was provided, I don't know, by Experian or Equifax, I'd probably be a lot more comfortable because I know they manage. They. They know a ton of stuff about me already and, and like, none of that's going anywhere. It shouldn't go to.

Speaker A: Yeah, I mean, Equifax had some problems a few years ago, but fair enough, I understand the concept. Right, yeah. I'm just trying to think, where does artificial intelligence fit in here? Do you know what I mean? Again, on both sides. So we already talked a little bit about how AI, uh, maybe can enable some fraud. Uh, I'm, um, dubious, to be fair. Most people don't know how to use their vcr, which is an old reference, but you know what I mean, they wouldn't even know how to program it. So the fraudster is going to have to spend a lot of time figuring out how to use Mid Journey to be, to be able to create fraud. Unless, like, that's their main business. But, but let's say that it still enables it. But on the flip side, how is this emerging technology? Because artificial intelligence, every day it just gets smarter and smarter and smarter. Um, and while I agree with you that I personally think it'll never replace humans because it'll never have intuition and I'm willing to die on that hill actually. Um, but it's still going to get used. I use it every day for a bunch of different things. Where does it fit in here? Particularly when it's meant to, um, engender trust.

Speaker B: I think you're right in that the average consumer probably doesn't have a clue what Mid Journey is, for instance. But to be fair, I don't really need to because I literally go into Gemini or I go into ChatGPT and I just, I literally say, create me a picture of a 3. No, I take a picture of my 3 series BMW for instance, and I say take that as the base image. Now create me an 8K image with a dent in the, uh, front driver's side door. Um, and make sure the number plate is my car registration. I give it my car registration plate. Like literally, like, literally do it on ChatGPT or. Yeah, I know, and I'm not kidding, like in 20 seconds, the image is there. Now, to most, to most claims handlers that are not using technology to validate if that's a AI generated image or it's another image. No, and, and then if you take a screenshot of it, it's really hard to kind of take any of the metadata from it. It's so easy to be able to do that so that the amount of fraud that as an industry we're seeing as a result of people using very, you know, very readily accept, um, available, um, applications, apps, um, to create false doctors, invoices, you know, you know, I could say I was out in Morocco. No disrespect to Morocco. Um, and I got taken into a hospital and I ended up having 2x rays and ended up having a pint of blood. And by the way, here, here's the invoice for it. You can get that kind of stuff super easy.

Speaker A: I just can't be bothered. But yeah, I get it, you can do it. I just cannot be bothered with this. But go ahead because people will do it.

Speaker B: But that's, that's down to your personality. Right? I, I just. Slight off tangent. I, um, I heard a stat once where they said if you, if you look at a group of people in a room, 40%, no, 30% of them are hardwired to, to always look for a way to, to defraud somebody. Wow, 30% would never do it. That's just not, that's not in, in their morals, that's not in their ethics. They're just not going to do it. But there's four, there's this 40% in the middle. That's my people. That will change depending on the circumstances. Right now they probably never have done it, but all of a sudden this morning they got a 2,000 pound utility bill and by chance the dog knocked the TV over. It was a 42 inch TV and here's my chance. I'm thinking, okay, I need some cash. Can I say it was an 85 inch TV? Um, I'm going to need a receipt for an 85 inch TV. I know John 2 Doors down just recently bought an 85 inch TV. Let me go ask him if he'll very nicely let me borrow a copy of, of his 85 inch TV invoice.

Speaker A: Yeah.

Speaker B: There's no name and address on the invoice, so like, yeah, so it's that 40%. And how do you, how do you detect if at uh, this moment in time they're likely to do something that's not normal behavior for them, but it's acceptable behavior for them right now. And that's the chat. The 30% that are hardwired don't have any issues with doing dishonest stuff. Um, you can profile those people because they typically fit a particular type of segment.

Speaker A: Um, yeah. And they probably drive a certain type of car and they probably live in a certain neighborhood or they have a certain type of house or something like that. Yeah. Although to be fair, you could have a very wealthy person who's committed a ton of fraud. You know what I mean? I don't think it's an income thing at all. You're right. I think it's a personality thing.

Speaker B: You know, I was, I was chatting to a high net worth insurer, uh, um, a few months ago and they were saying, you do realize our client base is very high net worth individuals. They don't commit fraud. And I'm going, I don't know what. Yeah, they do live on. But they definitely do, they definitely do.

Speaker A: As a matter of fact, some of them created all that wealth through fraudulent methods and they're going to fraud you as well along the way.

Speaker B: And the difference is you probably don't want to take them down the process because they're probably paying you so much money. Yeah, as a premium, you're prepared to let a lot of that stuff just slide. But they do commit fraud.

Speaker A: So talk to me again, talk to me again about how AI helps us solve this problem. If it does, if we can at all. Like where and how.

Speaker B: Um, I think so. There's generally, I'm just going to step back a level first. So there's generally two ways that insurance companies and actually most industries I try and identify fraud. One is to use all of the data that you have within your organization and you create profiles just the way that you've just described. You drive a particular type of car in a particular type of area, uh, your particular age band, whatever, and then you apply that profiling to your data set and then say if anybody comes in with a claim or wants to buy a policy and they fit this profile, you need to treat them slightly differently to people that don't fit that profile. So that's one way, but it's super biased. You can't help. You know all the AI is doing there is just allowing you to segment more tightly and speed up, uh, the definition of those segments.

Speaker A: Yeah, it's implementing my bias at scale.

Speaker B: Absolutely. That's literally all it's doing. It's not introducing any in new data because by the sheer premise of how AI works, it needs data to feed it to be able to create those solutions. The other side of that is to look at the evidence. So by evidence, um, um, is that ah, proof of Naaclem's bonus that they provided, is that genuine? Can we ring up their previous insurance company or get an email from them confirming they did have four years in their claims bonus or is that image of the damaged car they've sent, is that genuine? Can we get a ah, vehicle inspector to go out and look at the car? Can we drop it into one of our preferred Garages get them to verify that car was there or has that image appeared on Google somewhere? So you can use AI to speed up, uh, that kind of process now using companies like Camcom Tractable to kind of go, does that damage look like it's genuine damage from the car? Was that an image that's been manipulated? And that's kind of what. And all AI is doing is literally just speeding that process up. Um, I guess the big difference between kind of where clear speed falls into that space is we go to the source. So my view is, you know, if, if you, if you haven't given me accurate information. Um, so as an example, you know, I can put a telematics box in your car, I can put an app on your phone to determine your driving behavior. But, you know, if you regularly break the speed limit, you know, if you're using the vehicle as an Uber vehicle.

Speaker A: Yeah.

Speaker B: You know, if you're regularly using your mobile phone to check social media or messages while you're driving. When I'm driving, your mind knows that. Yeah. Rather than tracking you, why don't I just simply ask you those questions? As long as when you answer yes or no, I can trust the responses. That's key. Yeah. If I can trust those responses, I don't really need to track you. I can just, uh, ask you those questions equally. If you send me evidence of your damaged car and I ask you, you know, have you manipulated the data or the image that you've sent through to me? If you say no, and I can trust that. No, why do I need to even look at the image other than to kind of estimate the cost? But do I need to. Do I need to look at the metadata in the image? Now? If. If it flags up as a potential risk, absolutely. I need to go do that. More intensive check. But if you flag green, why. Why do I need to waste my time doing that? And it's very similar as an analogy to the metal detector at the airport. So if we didn't have those metal detectors, uh, everybody would have to be patted down in order to see if there's anything on me that I shouldn't have on me. The reason we use the metal detector is it's a way to kind of go, okay, walk through this little frame. If you've got metal on you, it's going to be red. And then a human expert, the security expert, guy or girl is going to have to pat you down. But if it flags green, I can just carry on. I don't need to be patted down now. If you take the same example of, uh, profiling, you're kind of going, if you look a certain way and you're a certain age, you have to walk through the metal detector. But if you don't, you can walk around the edge now. Yeah, we wouldn't accept that. As a society, we wouldn't accept that.

Speaker A: No, because if you had, if you were the guy that they had to make walk through the metal detector, you'd be like, what is it about me? And why does that guy just get to go the easy way? This would be a nightmare.

Speaker B: I travel a lot, and honestly, the number of times I get told, oh, you've been randomly selected, Please look at me.

Speaker A: You think I don't get randomly selected every single time I fly?

Speaker B: Literally, I get it all the time. And I just kind of go, fair enough. But when you look around, you kind

Speaker A: of go, yeah, that guy doesn't look anything like me. And he's just going right through.

Speaker B: But that feels unfair, right? And that, again, kind of comes back to, trust me. It's like, why are you. And because it fits the profile. I get it. You know? So, um, so I think that's the difference. The kind of profiling or the evidence checking. And by the way, on the evidence checking bit, the problem is the fraudsters and the general consumers that get tempted. They have access to technology that's way more advanced, way earlier than corporates get. So as an insurance company, in order for me to implement a solution that can validate whether an image has been manipulated, might take me 4, 6, 18 months to get through my procurement process and get validated, to be allowed for me to be able to allow to use it. That's for me to get a hold of version 2 or version 3. Problem is, during that time that software manufacturers on version 10 and the consumer is getting access to version 10. So one of the problems as an industry we have is how do we stay current and up to date? It's really challenging because there's new technologies appearing at a crazy pace, way faster than we as an industry can kind of keep pace with. So you have to change the rules. The type of fraud that's being committed today, has it evolved quite dramatically. But we're still using yesterday's technologies to try and solve today's problem. And that's really difficult to do. I think that's the value that we add as ClearSpeed. We kind of don't look at. We don't look at either end. We go to the source. And that applies the same way. Let's say you've just finished your degree. And I'm saying that because I'm on a university campus, loads of students that are graduating, so. So typically as a university they are used to putting the dissertation through an anti plagiarism tool to see if people have cut and pasted content. Okay, yeah, that's fine. But if I use AI to create some part of that content, that's way harder to detect because it probably is genuinely created. Or what if I paid somebody in Malaysia 500 to write my dissertation for me? How are you going to detect that? So to me that's a today problem, which yesterday's tools don't solve. But the bottom line is, you know, you did that. So if I can just ask you that question and I can determine whether. Actually, hang on a second. Maybe I need you to come in and answer a few questions about content in your, in your dissertation. I can't do that with everybody because I think in the UK we have 800,000 graduates every year.

Speaker A: Yeah, it doesn't scale.

Speaker B: It doesn't scale. So you have to be able to do it really, really quickly and be able to allow, um, the people that are genuinely not doing anything wrong to be able to get through that funnel super efficiently, virtually no friction, and just be able to get on with their life. Which allows you to then use your very limited resources to focus on, um, the needles in the haystack rather than the hay itself.

Speaker A: So before I let you go, where is all this going in your mind? Do you know what I mean? Like, what is it going to look like three to seven years from now? Because the fraudsters are going to keep frauding. And like you said, I know this because I can go down to Fortune Town, which is, I don't know what the equivalent is in the uk, but I can go down to Fortune Town road now and build the most sophisticated computer for like a thousand dollars. And then in a year from now I can go back and build something that's twice as powerful, twice as fast and has much better software on it. I can just keep getting better software.

Speaker B: Right?

Speaker A: But if I'm competing with you, if I'm trying to fraud against you and use software to do it, or sophisticated computing to do it, I'm already a year ahead of you when I buy mine because you're still in the midst of, like you said, a six to 18 month procurement process. So where are we going? And it's an arms race. And it's always been an arms race. Right, but like, where is it Going

Speaker B: I think the key difference here is that a, we're not really an AI tool. We're, we're a neuroscience based tool. So our technology works with um, in the way, the way that your brain responds to questions and you, you and it happens so quickly you're not in control of it. So when you're asked a question, your brain calculates the best way to respond to that question. Right now at this moment in time we may get a different signal if we asked you three days earlier or three days later for instance, but it's in this moment in time for this particular scenario. So um, we're the only company in the world that's able to do this today. Doesn't mean to say somebody else isn't going to come along and be able to replicate it, but we're in a pretty unique position. So we're used in very secure, very secure government, defense, military operations. So that technology is evolving because people that are trying to do really bad stuff are ah, really sophisticated when it's life or death situations for sure. And uh, we're using that technology in situations that are not life or death in, in, in places like insurance. But it's the same technology now. We, we just, we happen to have an edge and we, we're just going to get better and faster and, and everything else. Now we're virtually at ah, like sub second response now already. So I don't know how much faster we, we need to, we need to be able to get.

Speaker A: But you'd be surprised. You can get much faster.

Speaker B: Yeah. Um, but in a commercial world, in a commercial like you know, if, if I, if I ask you a question, um, like you know, you say your vehicle's been stolen and ask your question, do you know where the vehicle is now? You say no, if I can tell the insurance company within a couple of seconds. You genuinely don't know where the vehicle is. Like how much quicker do I need that response? Yeah, not much quicker but, but um, I think so when, when what do you use?

Speaker A: Are you using like, like voice detection software to figure this out? Do you know what I mean?

Speaker B: It's not voice detection. Um, so basically when a question hits your, hits your brain, um, your brain tries to work out the best response in this scenario and there's a whole lot of signals that flying around yet kind of like same as the fight or flight type reaction is instantaneousness. Um, our technology uniquely identifies those universal voice characteristics that exist in every human on the planet. Um, so you can't, you can't it, you know, it doesn't matter what culture you're from, doesn't matter what part of the world you're from, doesn't matter what language you speak, what accent you have, because we're not in, we don't look for those.

Speaker A: Yeah, you're not interested in that.

Speaker B: You're looking for these unique signals and uniquely we can identify them and measure them. Um, so then all that needs to change is the questions dependent on the risk. So the same technology underneath is, in principle, it's doing the same thing every time. It's looking for risk signals in your response. But if I want to know if you've taken performance enhancing drugs, I just ask that question. Have you taken performance enhancing drugs in the last 30 days? You say no, and I'll be able to tell you. Actually you have. I can't what drug you've taken.

Speaker A: Right.

Speaker B: I can't tell you when you took it in the last 30 days, but I can tell you you have. And then that therefore somebody needs to go have a go dig, dig a little bit deeper. But if we're saying you're clean, you don't need to do a blood test, you don't need to do a urine test because you're not going to find anything. Um, so that's the difference.

Speaker A: One more thing. Just because I'm super curious about this, this voice analytics thing, right. I just think it's super cool. Um, and you're right, the human brain is just like this fantastic machine that reacts in real time. And again, I think this is just another thing that most humans don't understand at all. Um, but if I'm trained to understand what the analytics are looking for when I'm lying, can I beat that system? If I know what those little things are that it's looking for, and if I know I'm going to be questioned. No.

Speaker B: Right. So in terms of you knowing you're going to be questioned. So we, we even tell people in advance what the questions are.

Speaker A: Yeah.

Speaker B: So actually it makes no difference. Doesn't help them in.

Speaker A: Go ahead.

Speaker B: Um, that. Can you train yourself to fool the technology? No, because you can't control the particular signals we look for. You can't control the way that they, they are generated and the way that they are emitted in your voice. So it. We're not, and we're not a lie detection. We're not saying if somebody.

Speaker A: No, I understand.

Speaker B: We're not, we're not saying. It's not like a polygraph talk. If your stress levels have changed or your tones, change. It, it's not, it's none of those. So, um, currently, can you train yourself to get around it? No, um, we, um, we can detect if it's an AI generated voice. Because that's probably another question that's in your head at the moment now.

Speaker A: Because I think, because again, like we can go. These AI generated voices are still terrible to the human.

Speaker B: They're getting better. They sound. No, some of them sound very accurate, but they don't contain these signals because they've been up.

Speaker A: Generated by a brain. Yeah.

Speaker B: Um, so yeah, we're using this technology to in very serious detecting terrorists, um, you know, life or death situations. So we, we can't, we can't have a solution that only works at like 70, 80, because that's not good enough. And those people that are trying to do really bad things in the world, they're pretty, they're pretty damn sophisticated. Yeah.

Speaker A: Yeah.

Speaker B: Aches are really high. Right. So, yeah. Um, so yeah, so I'm pretty confident that at least for the time being, you're not going to be able to get around this.

Speaker A: I'm sure we could do an entire show on voice analytics. I just, I love this topic because it's just the intersection of so many things that I love. I will let you go because I've kept you for really long. I hope you enjoyed this as much as I did.

Speaker B: This is awesome.

Speaker A: Um, Majid Rani, EVP of Insurance at ClearSpeed. This was awesome. You've got to come back on the show. Thank you so much for doing this.

Speaker B: My absolute pleasure. Thank you. Thank you for inviting me on the show.