In this interview, Startuprad.io talks to our first startup from the Czech Republic. We are very happy that they made it to LendIT in London, where we talked to them. We talk to Michael, who is Co-founder and CEO at ThreatMark.
The startup ThreatMark helps short term lenders and banks to minimize the default risk on loans with online application forms. They make the forecast, which loan is going to repay in full, better. But not only that, but they also help the companies they are working with to identify online fraud in credit applications.
Tune in to learn more
00:06: Announcer: Welcome to startuprad.io. Your podcast and YouTube blog covering the German startup scene with news interviews and live events.
00:20: Joe: Hello and welcome everybody. This is Joe from startuprad.io. Your startup podcast and YouTube blog from Germany. I’m still here at Landed in lovely London. And I do have another startup here with me. But you’re not from London. By the way, welcome.
00:35: Michael: Thanks.
00:36: Joe: Can you introduce yourself to our viewers?
00:39: Michael: Sure. My name is Michael Dresdner. I am the CEO of Treathmark. I’m coming from the Czech Republic from a lovely city called Bernard.
00:47: Joe: Very nice, welcome. I was just down here at your booth and you told me what your company actually does. And I find it pretty fancy zone. Basically, you’re reducing risk in loan applications. What are you actually doing and how you do it?
01:07: Michael: Okay, so actually this is our second product, which is based on the same technology as the first one. But this second one is really designed for an online lending business. It can basically predict if the user will go to default or also identify fraudster only based on behavioral biometric profiling. So we look at the things such as how users are typing on a keyboard, how they operate the mouse. And what is the interaction pattern with the form elements, for example, when filling in the application? So based on those measurements we can tell if the user will default at the end or not.
01:44: Joe: Actually I never paid any attention to how I fill out forms that will change from now on. And I was wondering how do you actually work with that? Is there like a basic rule like everybody like me who are doing completely crazy, will default like yesterday? Or is it more nuanced?
02:08: Michael: Yeah, it’s a much more complex of course. It’s based on machine learning and it is using big data. So we need to collect, we first need to collect a lot of data about the behavior of the users. We see thousands of applications. Then we are provided by some feedback. So we know which behavior of what users went to default. And what were the users that really paid the loan back? And based on let’s say a comparison of the current users’ behavior to the groups of behavior that we have seen previously. We can tell using this complex machine learning models if the user will actually go to default or not. So it’s not easily fooled by any let’s say things that you can really modify by your will. It’s something that is really deep inside your brain and you cannot really modify it.
03:00: Joe: It’s like subconscious behavior.
03:02: Michael: Exactly.
03:03: Joe: Okay. And what I also found interesting when we talk before, you said there’s like no blueprint. Like always this works, always that works. You have to adjust it to your bank because you’re actually looking for banks as clients.
03:19: Michael: Yeah. We are actually looking at those. Let’s say short term lending providers. Because these are the most vulnerable. But of course, even banks can benefit from this solution. But as you said, it’s not easily done in a way that we are looking at some specific behavior. We are using machine learning to adjust automatically what is actually the behavior of the users for that specific platform. So basically it needs to be learned again and again for each new application or each new lender for example. Because their interfaces are different and the system will see different behavior on different elements per user.
04:00: Joe: I’m wondering how many clients you already have. And do you see some like basic similar behaviors? Because Nisha one said if the order and chaos ever get into battle, chaos will win because it’s better organized.
04:17: Michael: Yeah. Like the solution from Threat Mark is currently used by around 15 banks and some other, let’s say, lending loan providers. But this new technology based on, let’s say behavioral biometrics only, it’s only used by few customers right now. And we are running a few PLCs. So what we see for now is that there are some common behavior patterns of a fraudster for example. So you can easily tell that it’s fraudster when he really knows how to go through the whole application. So we measured timings between let’s say the filling of different form fields. And by that, you can tell that it’s really a tech-savvy person that knows what they want to achieve. Quite often they even have a prepopulated synthetic identity that they are using simply by copy-pasting to the form fields.
05:10: Joe: I see. So basically if there’s someone who’s doing it as a fraud, he not does one but like 20 applications. And he usually just copy and paste, very simplified.
05:23: Michael: They can even only try to do let’s say one application. But even for the one application, they act very differently to what the general public can do.
05:36: Joe: Okay. And do you like also match it with data like women, men, age groups and stuff like that?
05:42: Michael: We try not to because of GDPR. So it wouldn’t be a nice solution if we can really match the gender or age or something like that. But of course it might be possible, but we don’t focus on that field.
05:57: Joe: Okay. Because I just want to ask you one last question. Who’s the most trustworthy? So, but apparently, you cannot do yet. We’ll have down here in the show notes, the link to your company website to personal LinkedIn profiles so everybody who is interested can reach out to you.
06:14: Michael: Okay, great. Let’s reach out to me. And it’s not only about lending, so we can help you to achieve, but I would also say any user identity verification in online channels and cyber threat detection.
06:27: Joe: Also like insurance?
06:28: Michael: Not like insurance. It depends, like if it’s done in an online environment then for sure. But what I meant is, for example, if you’re running some travel industry, some e-commerce platform or let’s says Bitcoin exchange. All of that are targets for fraudsters. And we can help to protect against those types of people.
06:50: Joe: Great. Thank you very much.
06:52: Michael: Yeah, thank you.
06:58: Announcer: That’s all folks. Find more news, streams, events, and interviews at www.startuprad.io. Remember sharing is caring.