FEATURING: Henry LeGard, Founder and CEO of Verisoul
In this episode, I speak with Founder & CEO of Verisoul, Henry LeGard.
Henry shares the founding story of Verisoul and the impact generative AI and large language models (LLMs) are having on online retailers. He discusses the ways e-commerce platforms faced with bot attacks can leverage Verisoul to level the playing field against sophisticated attacks. He also discusses how their 360 degree view of digital customers ensures users are real, unique, and trusted to stop fraud and fake users.
RESOURCES:
Connecting with Henry LeGard
LinkedIn: https://www.linkedin.com/in/henrylegard/
Verisoul: https://www.verisoul.ai/get-started-with-verisoul
Companies & Resources Discussed
Verisoul: https://www.verisoul.ai/
Neustar: https://www.home.neustar/
TransUnion: https://www.transunion.com/
Persona: https://withpersona.com/
YouTube IDV Spoofer: https://www.youtube.com/@bssports6805
Article about WormGPT: https://slashnext.com/blog/wormgpt-the-generative-ai-tool-cybercriminals-are-using-to-launch-business-email-compromise-attacks/
Article about FraudGPT: https://hackernoon.com/what-is-fraudgpt
Disposable Mail: https://www.disposablemail.com/
ARXIV CAPTCHA: https://arxiv.org/abs/2307.12108
FULL EPISODE TRANSCRIPT
Steve Craig:
Welcome to the PEAK IDV EXECUTIVE SERIES video podcast, where I speak with executives, leaders, founders, and change makers in the digital identity space. I'm your host, Steve Craig, Founder and Chief Enablement Officer at PEAK IDV. For our audience, this is a video first series. So if you're enjoying the audio version, please check out the video recording on executive series.peakidv.com where you can watch this full episode. There'll be a transcript. Any of the links or resources that we discussed today will be posted there. You can also watch the full video now on YouTube.
I'm very excited to introduce today's guest. He is Mr. Henry LeGard, founder and CEO of Verisoul.
Verisoul helps businesses and communities stop fake users like bots, multi accounters, fraudsters. Henry founded Verisoul in June of 2022, along with Raine Scott and Niel Ketkar. Prior to Verisoul, Henry was Director of Corporate Strategy at Nuestar.
Welcome, Henry. I'm thrilled to have you on the podcast today.
Henry Legard:
Thanks, Steve. Thrilled to be here. Really appreciate all the work you do for the identity and fraud community and super excited to meet you and the audience.
Steve: Thank you very much. Well, let's just dive right in. I'd love to hear what your 30-second elevator pitch sounds like right now. If you were in the elevator with me telling me all about Verisoul.
Henry: Absolutely. Verisoul is the easiest to use platform that stops fake users. We help businesses stop bots and AI, duplicate accounts, and fraud.
It's the only integration our customers need to stop these fake users. Handles all of the data collection, machine learning models, and decisioning for them. And it works directly out of the box. We're helping companies in the gaming, e-commerce, social dating spaces.
Steve: Very interesting. And with your time at Neustar, how did that influence your decision to pursue this particular market and to launch?
Henry: Yeah, it's a great question, Steve. So the time at Neustar helped me understand how the fraud and identity space has evolved. And when I think about the space and where it's headed, I generally think about kind of fraud and identity in two hemispheres. The first hemisphere is answering the question, who is this user?
A lot of the guests you have on Steve are kind of identity verification type companies. And most of them are answering this question. They're helping regulated companies like banks, fintechs, insurance, gambling, identify if the user is who they say they are. Now that's kind of only one hemisphere of the equation.
There's an entire other half that we see and we se rve. And that half is trying to answer a totally different question, which is, is this user real, unique and trusted? It's companies like gaming, social, dating, e-commerce that don't necessarily need to know you're Steve Craig, but they need to know that you're a real human, not a bot, that you're a unique account that's not abusing their systems and that you're trusted.
And so what we do at Verisoul is really solving that second hemisphere and how my time a Neustar helped evolve this vision is this. So, Neustar, you know, we ended up selling the company to TransUnion. And so those two companies generally are selling to highly regulated industries that need to know who the user is and therefore are comfortable paying a pretty penny to do so.
The customers we're serving aren't experts in fraud, they're less competent in this space. They haven't spent time in broad like banking and insurance companies have. And those companies have huge teams working on this problem. And Neustar and TransUnion, they sell point solutions that kind of fit right in and those companies build models around it.
Now, what we realized, or what I realized that Neustar was that there's a whole set of underserved verticals that don't have the background in fraud. They don't have the competency. Or that kind of need to understand how it works and really just need a platform that's super easy to use and does it all for them.
And so what Verisoul is, is it, you know, 15 minute integration that gives you banking grade fraud prevention, for companies that traditionally, don't, you know, don't build these models in house. And so that's kind of where the insight came and how we've transferred that into that, what we're doing at Verisoul.
Steve: I like how you frame those two hemispheres, you know, is this person who they claim to be, he or she claimed to be, and then does this person even exist or are they a bot or are they a fraudster? Can you trust them? I think. The unregulated industry, sometimes they don't care about that depth of, is this person really who they're claiming to be?
They just want to make sure that they're not letting in, sort of the automated scripts and things like that. So I like how you frame that.
Henry: Sometimes even those customers don't want to know who the user is for compliance or privacy reasons. And so some of the customers we're serving are actually switching away from traditional KYC and identity document verification to our technology because they don't want to invade user privacy or have that compliance risk of all of that PII faces and, and IDs in their databases. And so there's kind of a user privacy and friction benefit of what we do versus, um, other hemisphere providers.
Steve: Yeah, that's a really good point because the privacy regulation is, is only getting tighter and tighter and really the, the best way to comply is not to collect certain data, like, and you're not having to worry about purging the data and retention policies. It's like, you're not even collecting it. That's great.
Can you share a little bit more about your team? You worked at Neustar and then I noticed, um, your co founders have really interesting backgrounds.
Henry: Yeah, absolutely. So. Our co-founding team is, is three total people. So me and Raine and Niel, as you mentioned, prior to Verisoul, um, Raine has worked primarily in the startup space, but he actually suffered this exact fake user problem before.
So he was the first product hire at a FinTech that was aiming to start a neobank for kind of underprivileged and underserved communities. And they suffered from this exact problem. 80% of their signups, Steve, were fake, completely fraudulent users. Those were people that passed their traditional identity verification processes, and all of this is because they were giving away at a referral, a $20 promotion.
And that, you know, giving away money on the internet is one of the hardest things you can do, as we always say. And so Raine literally suffered this exact problem where he tried tons of different kind of point solution vendors. He ended up trying to stack all of these different technologies, right. KYC identity verification, device risk, IP risk, etc., stacking all these point solutions realized,
Hey, it's really hard to integrate all of these.
Takes us a ton of time. And then we've got to go build our own custom fraud models on top of that. And it was a huge headache for him and his team, which had never dealt with this problem before. So that was another, you know, kind of, helpful in set of insights as we were founding Verisoul like, Hey, you know, teams that are product or engineering focused, they should be building what they do best and focusing on their user experience, their platform rather than spending all their time doing fraud prevention.
And then our head of technology, Niel, um, prior to Verisoul was working at CapitalOne. And a lot of what he was doing there was using customer activity and kind of intention plus identity information to generate predictions on what they needed to serve that user using AI.
So he's taking combining behavior plus identity and generating kind of work flows and predictions around that. That's exactly what we do here at Verisoul. We're looking at user behavior and information about that user to determine if they're real, unique and trusted. And then if not, or if yes, serving, helping customers serve them a different and the right workflow for them, you know, based on that predictive decision.
Steve: Incredible. It sounds like you've got a rockstar team of people who have been out there already solving this and now bringing it to the masses. Was June 2022 the right founding date? I saw that on your LinkedIn.
Henry: That is correct. Yeah, a little over a year. The three of us actually, before we started the company did some projects together, helping a friend of ours, actually, this is pretty crazy.
They were raffling off a ticket to space. So they literally bought tickets on blue origin, Jeff Bezos's rocket company. And they bought two tickets. They were raffling them off for free. And one of the lesser public known fact is that for a raffle, you actually have to make it free in the US and so anyone could enter for free and you also had to have mail in ballots.
And so one of their biggest challenges was because this is an incredibly high value. I mean, Steve, you would do basically anything for a ticket to space, right? Like these things sell for 500K and they were highly worried about fake users, bots, automated scripts, attacking their platform. And so before we even started Verisoul, we were kind of working together on, you know, ideating on our backgrounds and, and where we could, where we could start a company together, and then this problem kind of fell in our laps.
And it resonated perfectly with us because it's exactly the combination of, you know, what we're doing at Verisoul now, but also the backgrounds we had. And so. What we did was throw together kind of an MVP, you know, the first really early version of Verisoul to help them stop users on that rough. It was highly successful.
You know, we found about 8 to 9,000 bots and fake users that were attacking their platform in hopes of winning this really high value ticket to space. And so, you know, that was back in May 2022. So then we did, you know, took that. kind of challenge and founded Verisoul.
Steve: That's a, that's a great founding story.
And you start the company and it's been over a year. What are some of the milestones you've hit in this last year or the customer pilots or launches, things you're most proud of?
Henry: Yeah, it's a great question. So look, generally, when I think about the success of the company, it's really about how much success can we drive for our customers, right?
The whole goal of building a platform is that it creates value for people you work with and companies you work with. And I'm, you know, I'm excited to say that we've recently passed a huge product milestone for our customers delivering the next version of what we call our zero fake platform. And this is really the bread and butter foundation of Verisoul, which is 15 minute integration, real time data via dashboards and API, and then configurable decisions for those customers. So what this means for them is it allows them to go from zero to secure in 15 minutes. It's an incredibly powerful platform for our customers.
And then other the kind of translation of that is into milestones, right? So now. One of the best parts of my job is really interacting with our customers and hearing how much insight our platform is driving for them, right? It's companies that didn't know what percentage of their users reeled through good users and what percentage were fake bots, etc. now understand that and can truly build for the right set of users, right? So it really gives them product visibility and it also helps them create better user experiences by keeping out a lot of the bad guys.
From a numbers standpoint, Steve, you know, I can't right now share any customer names or numbers, just given the nature of the security business, we don't like to give away too much in terms of you know, giving fraudsters the roadmap on who to attack, who not to, etc.
But.Wee are moving, moving fast and, uh, plan to be growing the team. I think other thing I'll mention as far as milestones and none of this, I expect to have some pretty meaningful updates in the next couple of weeks that are in the work, so really excited about what's in store from the team, for our customers and for our product.
Steve: Totally understand that. You don't want to telegraph a specific logos, customer names, and then the fraud's like, Oh, you know, a challenge. Let's go try to nail that. Or then those that aren't using you become targets. Well, this audience can be very technical. Can you describe maybe a little bit more about what that 15 minute implementation looks like?
Where is it? SDK? Like what's the implementation journey look like for enterprise? And then what is the customer experience? If anything, like what do you even know?
Henry: For our business customers, right now, the integration after our zero fake platform is one SDK that works kind of wherever they, it's web, it's mobile, it's even a Unity engine. So we work with a lot of gaming companies that are built on unity gaming engine. So we have an SDK that plops in directly for them.
This is as easy as a Google analytics tag, right? A script tag style integration. And then, um, that automatically enables real time data via a Dashboard or that gets set up in about a minute. And so you can go from, from zero insight to full visibility into your user base in literally less than 15 minutes.
And then for customer, for our business customers that want to enable real time workflows. They can integrate via our API, which takes, you know, a couple hours, as any API integration does that gives them sub second API responses. And that enables them to create real time workflows for their users.
And so now to your second question, Steve, which is what happens to the actual users? Our view is that each ecosystem and each business we protect is completely different. And so the way we think about this, Steve, is we actually give the dials and the kind of read to our customers and we allow them to within our platform, build workflows for those users.
And what that might look like is completely frictionless. So for many of the customers we work with, no users ever, know Verisoul even running. For some of the, for some of our customers, um, we also, we also are kind of step up authentication. Meaning as a user is deemed to be suspicious or likely risky, we offer within our platform components that our customers can serve their users to do step up kind of verification.
One of the core step up methods we've built is what we call zero KYC. This is similar to a proof of human, which, generally we're looking at two main things.
The first is liveness of the user. So we leverage voice and face, real time video and audio capture to detect aliveness. And we also use those same signals to do uniqueness. So we're trying to determine, are you a unique user or not? Now what's really powerful about this, Steve, is a couple of things. The first is how accurate it is.
One of the customers we work with literally just switched away from Persona, which is an identity verification provider to us because they were getting rocked by deep fakes, video injections, kind of 3d images that people were just injecting into the video stream and it was bypassing their security layer and that was, you know, costing them that cost them customers, literal, you know, they're, they're a B2B platform and it cost them business customers because fraudsters were getting through.
The other thing is because those traditional IDV providers use ID documents, it's a lot of friction for users. And so what we do is we remove that there's no identity. There's 100 percent anonymity built in. There's no identity document, no name capture. And in fact, all of the biometric data we capture is actually hashed and stored only in a like numerical vector format, which is totally privacy preserving and anonymous.
So for ecosystems that don't want the friction of identity documents and they want better, more accurate security. They're switching to our Zero KYC.
Steve: Fascinating. For Zero KYC in some of these use cases you described, are these so detection activities specific to a single customer, or do you have like a network effect where if it's one customer seeing this attack, then maybe another customer in a different vertical even is getting the benefit of.
Henry: That's an awesome question, Steve. Yes, we have what we call our alliance, which is all of the customers. We have benefit from all of the intelligence we gather across the customer base views from a privacy perspective. We're never sharing user data. We're only sharing insights and decision information.
And so this is one of the things I learned when at Neustar and TransUnion, the power of a cross customer identity graph. And so underpinning our entire business is a graph that connects people, devices, networks, locations, and behaviors. And so what that allows us to do is use insights on fraudsters from one customer ecosystem and propagate that to all of the customers so that they benefit and that creates a flywheel. So, you know, with each new decision we have, or we get that information becomes relevant and powerful for other customers. And so that's core to Verisoul and core of what, you know, creates an ever improving flywheel of intelligence.
And it's another reason that we recommend that, you know, businesses don't try to build this themselves, right? A lot of what we're going up against in sales is actually just customers that have built their own, you know, ad hoc scripts and stuff like that, that are getting a lot of false positives, but don't benefit from this network effect, right? And, you know, when you think about fraud as an arms race, what we always say is, you know, if it's an arms race, you don't want to be fighting it alone. Right. And so there's value in the alliance.
Steve: For that alliance, is that something that is automatic or is it an opt in?
Do customers have choice whether or not they contribute?
Henry: It's a great question. Right now it is, uh, automatic, uh, or maybe opt out, I would say. Opt-in by default, and we've really, most customers are kind of excited about the prospect of that, um, of that additional intelligence and benefit. There are, you know, longer term, I definitely see, you know, larger enterprise clients likely not interested in sharing proprietary information like this across the network.
And so, I think down the road, there's a lot of opportunity to build, I would say self contained instances of Verisoul ZeroFake and deploy that either on premises or in kind of a containerized way for larger enterprise clients, but even then there's, I don't know how much, um, you're spending following the clean room type development, Steve, but there's a whole kind of sector of, of technology developing around clean rooms that will allow, that will allow two parties to share insights on their data without ever exposing the true information that exists.
And so they can actually, using homomorphic encryption and other technologies. Two large enterprises can benefit from each other's data and insights without ever kind of sharing any user data. And so there's definitely desire and appetite for large enterprises to get the benefit of this kind of Alliance-like framework.
And there's a lot of technology that's adapting pretty quickly to meet that demand.
Steve: Yeah, I've certainly encountered large enterprises that get in this mindset where they don't want to share data, they won't want to focus only on their ecosystem, and then they lose the benefits of that advanced awareness of these fraud attacks.
And we all know the fraudsters coordinate, they go on dark web forums, or maybe even just as basic as Reddit and say, Hey, this is how you bypass the system. And it's really important that companies, both solution providers and the practitioners work together. And if they can do that in a privacy preserving way where they don't have to be concerned about oversharing their customer's sense of details, I think it's a powerful way to fight back and this sort of turn, turn the tide.
You mentioned something interesting, just to ask. Just now about the status quo or companies might start to build this themselves. And then they quickly realized that it's a tough problem to solve. Many companies also think about a single account as being, well, this is an email address or a phone number.
I'm going to click a link to verify, or I'm going to do a one-time password. I was looking at your, some of your demos and I see you have some intelligence around email and like email information. Can you share more about how that works in conjunction with some of the other pieces of the platform.
Henry:
Yeah, absolutely. So our email intelligence, you know, I would say email intelligence on the whole is kind of a table stakes solution. I wouldn't say that alone it's that valuable or it's not your, it's not a silver bullet. Let me phrase it that way. You need it as one component of a 360 degree view of your user, but you shouldn't rely on it only.
And really what we're doing when we're building our, or when we're analyzing emails is kind of three things. The first is, is it deliverable, right? This is super table stakes. You know, I wouldn't not nothing differentiated there. Number two is, is it disposable, right? Is this a known kind of fake or disposable email?
And when I, when I say that, what I mean is. There are these sites like disposalmail.com or whatever, where I can go at any time and pull an email off their shelf and basically use that to create a fake account. Now what that allows me to do is create infinite accounts on any given platform, right?
Because these emails are not tied together, they're not mine and I can do it in one second. It takes no time. And then on top of that, I could build a bot that goes and does that automatically. Right. So the kind of scale I can use disposable email for is quite high. So we're checking whether that email is a known disposable domain or likely sake.
The last thing we do with an email intelligence, and this is what I would where a lot of the proprietary technology is Steve and where most solutions don't really cover this is we're doing the deduplication. And so we're running our own models on top of the entire email base to look at, to graph out the emails and figure out how, what's the linkage between any two emails and how strong does that link.
And by doing that, we can analyze how likely it is that the users created multiple accounts. And so this is a unique offering in market. The other thing that makes, you know, our email intelligence unique is its free. We actually just give it away to companies that, you know, let's say they want, um, they want to start with this as a way to understand, Hey, what does Verisoul have to offer, but want to do it for free.
You know, we actually offer this to customers for free, um, to get them started if they don't want to leverage the entire platform. The other thing is just how easy it is to use, right? It's super simple to integrate, receive responses and that makes it valuable in the market. But again, the way we think about fake users is really you need a 360 degree view of that user.
You can't just have one point solution because if you do fraudsters will quickly adapt to the next kind of think about it like a dam that beavers are building. Right? You know, you put a branch in the water, it just keeps moving around. And so you need a 360 degree view of that user to build all of the detection around them to really prevent them from evading your defenses. So email's a tool in the tool belt, but not a silver bullet.
Steve: Yeah. Yeah. Well, going through these experiences as a consumer. They're all different. Sometimes you are clicking a link. Sometimes they send you a number. They're just extra friction for us as good consumers and without that additional intelligence of fraudsters automating past those.
I mean, they're probably verifying the link immediately because they created a script for that. And I think about the, like this automation of fraud and like the scale of fraud. And in this last year, as you've been working on Verisoul, we've seen large language models, we've seen ChatGPT go mainstream.
It's really given bad actors like extra horsepower to be able to defraud people at scale and whether that's for phishing or trying different combinations of attacks.
What are some of the things that you've seen just, I don't know, in the last six months that you've gone like, wow, and that you've then had to?
Henry:
Yeah, it's a great question. Look, this, this is a step change. It's a game changer for fraudsters. It's like going from a bike to a car. I mean, it's crazy. It enables so many more fraudsters to create highly sophisticated bots, fake users, etc. The way we think about this is in kind of three main themes.
The first is better fraud. So, you know, if you think about a foreign fraudster, they might have had broken English or might not understand English. Let's say they're attacking a state or municipality citizens, right? They're likely to have pretty obviously broken English, etc. Now with the onset of AI tools, what was once broken or kind of incomplete, unnatural language is perfectly natural human sounding language.
That in its own is an incredible step change in terms of the ability to defraud users. The second is within this better broad umbrella is deepfake content that's highly accurate. And to give you an example, right? I kind of mentioned this earlier, but one of the, one of our customers switched away from their identity verification provider because of exactly this, anyone in the world can create highly realistic 3d deepfake content that can scam traditional identity verification providers.
In fact, there's a whole, you may have seen this guy, Steve, but there's an entire, there's a guy on YouTube that basically create videos on how to spoof all of these ID provider and he does it in five minutes. It's incredibly straightforward. The other kind of theme or the other, uh, last point within this better fraud umbrella is the speed and scale.
Now, anyone that has access to a computer, even without computer science or engineering programming backgrounds. Can build bots, fake accounts that are highly performant and highly scalable. The other emerging theme is what I would call tooling. I don't know if you've seen this, Steve, but there's an entire set of tools just for fraudsters called WormGPT, FraudGPT, etc.
These are instantiations of LLMs designed to build phishing sites, to write malicious malware scripts. It's incredible. You should take a look at these because it's quite informative as to how powerful some of the tools are that are now in the hands of any. So this is a step change in terms of the tools accessible and really what this what it does is it doesn't, it's not making the most sophisticated fraudsters 10 percent more sophisticated.
It's really making anyone go from 2 percent sophistication to 50, 60, 80, right? And that's where all the volume is. And then the last kind of theme that we think about is it's harder to tell who is human now. And there was a great research paper that was just published on archive ARXIV.
I don't know if you've seen that. It's kind of an open source journal publication platform. Elon tweeted it out. It's been big news, basically CAPTCHAs are, or sorry, bots are now better at solving CAPTCHAs than humans. And what I mean by that is they are more accurate and faster. And so it's better than, you know, bots can literally solve CAPTCHAs more accurately, faster and go undetected.
And that's across a range of CAPTCHAs. That means that traditional, you know, re CAPTCHA, ARCOS, CAPTCHA, etc., they really can't be trusted because. Fraudsters can just get right through them in fact, better and faster than your real users. And so what ends up happening is you get more user abandonment than you do fraud prevention.
And it's incredible. It's a complete reversal of the script here. Some of the details, you know, some of these captcha providers are, you know, creating friction of 80 seconds for users, at like the 75th percentile. That creates huge abandonment. So companies today are basically faced with an easy decision now, which is, do I want to add friction for only my real users and the fraudsters I'll get right through, or do I want to find another way and, you know, so that's kind of the, those are the three themes that have jumped out with respect to the onset of AI and how it's affecting fraud.
Steve: That's very insightful. And I hadn't heard of a few of those. I'll be sure to get those after this session and I'll post them as links. So anyone who's watching or listening could go check out those resources. Just something we see from fraud first, because there's no regulations for the fraudsters, right?
There's, they don't worry about, Oh, I'm breaking this law or that law. There's no tape or legal review. They just do it. And then they go, wow, that worked. And then they share it and it catches fire quickly in these communities. And then you as an enterprise, or you even as a consumer are getting the short end of that deal for sure.
The bot area of the CAPTCHAs, like we all hate CAPTCHAs, like whether it was trying to figure out the squiggly lines or doing math or the last decade, it's been like picking the images, like where's a stoplight and where's a van or those things, it's painful. And I could see how these technologies from the fraudsters are able to pick that out quicker.
When we think about bots though, whether it's buying concert tickets, like to the Taylor Swift show, or maybe you want to buy sneakers that are limited edition. Maybe it's technology like Nvidia cards or GPUs. This is problem for us transacting on the internet because we are all trying to put it on a keyboard or on our phone, and then everything's sold out because there's some script that's pulled all the goods before we can even refresh the page.
For those that are in the e-commerce world, how would they use Verisoul to stop these? Like what, what's a good kind of use case commercial in a way for, for stopping these, these challenges?
Henry: Yeah, that's a great question. I feel your pain, Steve. There's a restaurant here in Austin, Texas that only sits like six people.
And I've been wanting to go for since it opened, like literally a year. Every month I try to get on at, you know, midnight, Sunday night, the month, you know, the Sunday it opens and you never can't, you just can't beat him anymore. And what's really funny actually and let me just get there a couple of things and then I'll get to how we solve it.
But what's really funny actually is. You mentioned the kind of sneaker bot world, that space is so competitive that it's not even bot versus human now it's bot versus bot. And you really actually have to have a 90th percentile or better bot to even get a pair, right? So it's not even like humans aren't even in the equation anymore.
Like the, only the 10 percent best bots get sneakers. And so what these guys do. Is they will only sell, they'll build, you know, what they claim to be the fastest, best bots, and then they'll only sell a limited number of them that way, so people that buy them think, okay, I'm one of the few that's getting this kind of cream of the crop bot, but I'm sure that fraudsters has the slightly better bot that they use.
It's pretty insane. The reason I know all this is because we actually work with some of these guys, these people that are building bots an for sneaker platforms for games, etc. And the reason we do that is so we understand what are the techniques they're using.
It really gives us kind of an on the ground first principles perspective of what these guys are doing and how it's evading current systems. And that allows us to build AI models that can actually detect them in real time. And, you know, so the way customers use our technology is we detect bot like behavior based on how really fundamental behaviors like mouse movements, clicks, touches, etc.
And there's really no other way to explain it other than we work with fraudsters, we get their best bot and we get a bunch of human data and we train really large AI models on what a human looks like and what a bot looks like. And in real time, we can detect whether it's a human or a bot.
And that provides the insight our customers need to either add incremental friction that can actually stop a bot that they're not expecting. Or, allows you to, you know, that platform to kind of quarantine or ban that bot from actually making a purchase. So, you know, let's take a hypothetical for a sneaker example, you might detect that it's a bot and kind of infinitely quarantine it in some checkout page where they're never able to actually check out when you have certainty that they're bought that way, they think they're, you know, going to get something, but don't, or you could reverse the transaction later. If you detect that they're a likely fraudster.
Steve: Any area that has scarcity, right? If there's a limited supply of something, you need to be able to put these measures in place to be able to help your true real fans who want to go to your show or buy a product because they want to own it versus trying to resell it on eBay or some other marketplace, give them that, that power. Otherwise they just say, forget it. And then someday there's no customers left because it's all just bot interactions.
It's a really fascinating area. And it's for me personally, when the PlayStation 5 came out, I was trying to get one of those.
I tried everything Best Buy, Amazon. It finally took Sony putting in some process for if you were a PlayStation network member that you could get an invite to be in the queue, right? It was just so much inefficiency that went into just buying something that should have been a basic transaction, like the complexity.
Henry: It's incredible. And do you, did you get the PS5 or are you playing it actively now?
Steve: Eventually I did it like it took a long time and I got one and then immediately I did think about selling it, but I said, no, this is it. And now, now you can buy them. It's easy, but the amount of time that went into that from an industry standpoint, just that one asset.
And I know this happens with, with all sorts of limited release and this was the supply chain crunch and all that.
Henry: I'm expecting a similar, we're expecting a similar kind of velocity for the Apple's Vision OS or the Vision Pro? You know, I think that'll be an incredibly limited release.
I think the price for that object is, the demand for that object pretty inelastic. I think there's going to be a ton of bot behavior when that goes live, and that'll be something we'll be monitoring. But you're right. I mean, it's a turnoff for normal people. Right? How do you, like, why would I even go to Ticketmaster?
When I know I'm going to sit in a two hour queue and never get the product. It's like and an interesting example because they're basically a monopoly and, you know, a lot of monopolies to behavior. So we'll see what happens with them, but pretty interesting. It's definitely an interesting space to be around.
It's really interesting to work with the guys who are building bots to take, you know, to attack these sites and hear how they think about it and how they only sell a few of them and, and how they make their money. There's a whole black market world and, you know, they're not bad people.
They're, they're just trying to make ends meet by selling bots or, you know, buying merchandise, but it's to the detriment of all of us.
Steve: They're, they're entrepreneurs and they're finding a market gap and they're trying to play in that space. So it's fascinating, but we have to do more to stop it.
Well. Henry, we're, we're coming up on time. I have just like a few remaining questions to close out this episode.
One is around some beta features I saw in one of your demos around detecting generated AI writing, looking for harmful text and then logical lies. I hadn't heard that. Can you share more about how that fits into your broader platform and how you think about kind of generated media?
Henry: Yeah, absolutely. You know, so we talked, I talked about this a little bit, but. We really think about stopping fraud and fake users as a 360 degree view. And you kind of need everything and the more signal you have the better. And so one of the signals we're testing and playing around with is leveraging AI in two different ways.
The first is to detect LLMs or to detect text written by LLMs. And the second is to use AI to actually detect logic, logical inconsistencies, lies, things like that, that would be clues to potential fraudulent or fake behavior. Let me give you kind of two real world use cases where this is valuable.
We have companies in our set of customers that are survey and offer providers. Right? So they host surveys and users complete them. And those insights become powerful data points for enterprises. They're looking at markets and analyzing customer behavior, things like that. It's critical that those platforms have high integrity user bases, real, unique, trusted people.
And that they're not duplicates, not written by AI, etc. And so a lot of our, you know, initial beta features are really trying to detect how likely is it that someone is using AI to record responses. And so the three things we're doing, uh, one is detecting that AI, two is looking, what's really cool about using this AI is we can actually look at the logical history of that whole account over their lifespan.
And so he's, what I mean by that is. Did they ever answer a factual question differently in two different occasions? So let's just say, you know, I was born in, in Missouri, right? Every time I take a test, I'm probably going to be, have to answer something related to geography or place of birth or something like that, but you can ask it in a bunch of different ways.
And if you use AI to analyze, you can look at hundreds of different surveys, how they've responded to that and ever see if there's a logical inconsistency within that bunch. And that allows you to understand whether someone's kind of randomly making things up versus a diligent, real, trusted, unique person answering surveys, um, you know, that that'll benefit enterprises.
So that's kind of how we're using, that's one use case. The other is in what I would say, the gaming and maybe the social and community space where, you know, if you can leverage. Technology to faster understand, you know, missed disinformation, AI generated content, etc. The more your, you know, the more information your team has to, to better pursue those users.
So kind of another area we're exploring, I would say these are, you know, beta features, but you know, one of the things we like to do at Verisoul is just ship new products faster and get it in the hands of customers and get feedback on it. And then if it's interesting and valuable, then, you know we continue to productize and if not, you know, we, we move on.
And so we, we think one of the unique parts about Verisoul is just how fast we move, you know, there really are none of our competitors have anything like this.
Steve: Yeah, when I saw that, I thought, well, how cutting edge, because we're already seeing big platforms, like even LinkedIn or Microsoft embed the ability to have generated responses.
And it's polarizing because people want an authentic voice from individuals. And that's from like a content creation standpoint. But then when you think about these private interactions, whether they be with chatbots or enrollment forms when you're building a bot that is sophisticated enough to solve a CAPTCHA, you certainly can answer these other questions.
And they're like, how do you catch them? And those, those white, white lies, I guess, the little fibs that later, uh, turn into bigger problems. That's very powerful. Well, we're just at time now, uh, if you've seen any of these other episodes, Henry, you know, I like to end with a couple of questions around about you, you know, the person behind the profile, the person behind the company.
Besides building Verisoul and tackling fake accounts and bots and fraudsters, what other causes are important to you? Like, where do you spend your time outside of the business?
Henry: Yeah, it's a great question. So let me answer this in kind of two parts. Wi respect to causes, I think the one I'm most passionate about is financial literacy.
I think it's one of the most under taught subjects. And one of the things that can have the most leverage on underprivileged communities. And in fact, you know, I tutor financial literacy at homeless shelters, etc. And one of, if not the largest reason that most people have insecure housing is a lack of financial literacy because they never had the opportunity to learn it.
And it's not because anyone's not smart or really just, it's never taught in public schools for some reason, it's a shame, and it's really hard to understand as written. And so a lot of what. You know, in that space, a lot of, you know, how I think about is how can you distill it down to a few really easy to use and easy to implement principles that can help folks better understand how finances work, how to think about their finances, but also how to employ them in a way that allows them to gain security and housing and finance and finances.
And then from a personal perspective, one of the things that I love to do personally, just bring in big events. I'm a really social person and really love large groups. And to the sum of what I spend my time doing is planning large parties and events for friends and you know, things like, an adult ball pit party with a million ball pit balls that we shipped in from China o a multi-level party that, you know, goes through the decade by going at each level of a building.
Right. So things like that are, uh, stuff, how, how I get friends together and, and, uh, love really what makes one of the things I do that, or one of the things that makes me have the most fun is just seeing others having a ton of fun by way of you know, my hands. So that's something I like spending my time doing.
Steve: That sounds fantastic. Yeah. I like to host functions through PEAK IDV. So maybe we do a PEAK IDV Verisoul Austin based event and we just bring. Everyone's to it from the industry or not, and just have a good time. That's awesome.
Henry: That'd be really fun. I was super down to help out with that.
Steve: Awesome. Well, for, I guess the last question for you is what kind of conversations would you like to engage in from this podcast episode? Like who should reach out to you and how should they engage with Verisoul? Like, what are you looking for from the market?
Henry: Yeah, it's a great question. Maybe two things. The first is I love debating, I love when people are like, Hey, I think you're wrong and here's why. And it's, and having a really good kind of principles based discussion around that is super interesting to me. So I'd love for anyone to chat about things I said that they don't think are right and why, and have an engaged debate on, those topics.
Second would just be from customers or even people that have felt this problem and, you know, how it impacts them, what they've tried in terms of how to fix it. And, you know, maybe why they haven't been able to solve it so far and see if Verisoul is able to help fill that void. We're still a startup and obviously not everything's perfect, but you know, the more people can share their problems with us, the more we can adapt our platform to help them. So that's kind of the second category of, of conversations I'd love to have.
Steve: Excellent. Sounds great, Henry. I'm sure you'll find some people who watch this, who do want to have some, conversation, maybe arguments with you over some topics.
Henry: Hopefully, hopefully. Yeah.
Steve: We'll see if we could also find you some, uh, some customers and some partners through this engagement. Thank you so much for taking the time and spending this almost hour with me to talk through Verisoul. I'm really looking forward to watching your continued success and you grow and expand.
And I will, come out to Austin, we will host a party. So don't just stay in that jest. Thank you again. Awesome. Henry.
Henry: Thank you, Steve. Really appreciate this, you know, this time as well as what you do for the whole identity and broad community. I think it's a really powerful set of insights you helped bring to people.
So glad to be a part of. Thank you.
Steve: Appreciate it.
Fake User Prevention with Founder & CEO of Verisoul, Henry LeGard