Hi, Tamas; welcome to our Interview Series. Please tell us a little bit about your journey in the technology space. How did you start SEON?
Technically SEON originated when I met my friend Bence in University in 2016. We bonded quickly over a mutual interest in the cryptocurrency landscape, and then pooled our abilities and resources to start our crypto exchange for the CEE region. We were immediately targeted by attacks from fraudsters, so we looked for the best fraud prevention solution for our business.
In our search for the best fraud prevention solution, we tried various options on the market but eventually decided to create our own, tailored to our specific needs. Our solution proved so effective that we shifted our business model and began offering it to other companies. Today, after a successful $94 million Series B funding, SEON has come a long way from its origins as a startup founded by two ambitious university graduates.
What is SEON, and what are your core features? Which industries are currently benefiting from your products and solutions?
SEON is an end-to-end fincrime prevention platform that relies on data enrichment, device fingerprinting and machine learning to give companies the edge in the fight against fraud. We take an email address, phone number, or IP address and check the digital breadcrumbs their owners leave online, like profiles registered on social media sites or other digital platforms. We then run all this information through our Scoring Engine and Machine Learning solutions to give them recommendations about who to trust. We turn a handful of data points into hundreds to help our partners tell good customers from bad quickly and without hassle. SEON is already trusted globally and, especially by companies in the iGaming (Soft2Bet, LeoVegas) and online lending industries (ViaBill, Biller), but we also work with ecommerce sites (Lalamove) and neobanks (BuBank and Revolut), to name only a few.
Brands and customers are at an all-time high risk of bot attacks. Could you tell how bot attacks identify / exploit weaknesses and loopholes in an organization’s digital footprint?
Bot attacks are most often leveraged by fraudsters when scaling is part of their methodology. This scaling may mean turning incremental payouts for advertising affiliates into significant total sums by pushing valueless bot traffic through paid gateways. Other times, as with credential stuffing attacks where bad actors have purchased or stolen login/password combinations, fraudsters need to scale their bots so they can try as many of their stolen login credentials on as many domains as possible until one works and they gain unauthorized access to an online account. With these two examples – some of the most common ones we see – it is hard to say that they are exploiting a particular loophole other than the human one. That is, companies that spend less human resources on vetting their advertising affiliates, and people who don’t practice good password hygiene by ensuring each domain registration has different login details, are more likely to fall victim to fraudsters deploying huge armies of bots. From top to bottom, companies should be keeping their staff educated on digital interactions they should not trust, and those staff members should carry those lessons home with them into their personal computing.
Please tell us more about AML and how you have brought AI and Machine Learning capabilities to boost fraud prevention?
Preventing money laundering is a very technical process behind the scenes. However, it has been designed to be the largest legal measures taken towards stopping international organized crime and terrorism. At SEON we’ve always aimed to create a toolset that gives our partners insight into every transaction, order and customer. Adding anti-money laundering to the mix always seemed like a natural next step in helping our clients stop fincrime. By acquiring Complytron we brought a team with huge experience in this field on board, and our customers already see the benefits.
Machine learning is another beast entirely. AI is great at crunching numbers and recognizing patterns humans find hard to spot. When we have hundreds of data points in each transaction, using AI to help churn through the data is a no-brainer. The difficult side of things is transparency and trust. We strongly believe that our customers should be in control of whatever happens in SEON; after all, it’s their revenue at stake, so we didn’t take the route followed by a big chunk of the fraud prevention industry and just roll out a blackbox machine learning tool that doesn’t explain how it reaches its decisions. Both SEON’s whitebox and blackbox tools utilize explainable AI, which means our partners benefit from the performance and speed of AI but without the drawbacks of not understanding how decisions are made.
Could you tell us about the importance of modern device fingerprinting in the modern scenario dominated by mobile-first and app-centric brands?
Professional fraudsters have a huge range of tools at their disposal to try and trick companies into believing they’re honest customers. Using virtual devices, for example, a single PC simulating 10–20 different smartphones, is an extremely common option. Even if they don’t go that deep, fraudsters can choose from various browsers, plugins, and even proxies or VPNs to help them cheat money from businesses. The best way to counter these highly technical threats is by creating technical solutions that recognize them and stop them in their tracks. SEON’s device fingerprinting tools differ from other offerings on their market because they specifically focus on fraud tools. We’ve honed our solution to catch suspicious hardware configurations and recognize the noise fraud plugins create to spoof other device fingerprinting products. As a result, instead of seeing that a transaction was made from a two-year-old android smartphone, our customers see that it was made from a virtual device, running on a strange server-like computer. One thing’s for sure, more honest customers don’t have a reason to use a virtual device, so that’s a big red flag out of the gate. We can also help our partners separate privacy-conscious customers from fraudsters by recognizing privacy plugins and VPN connections. This is critical, because some privacy tools are, sadly, also being used by fraudsters. Being able to add more information about the full-range of browser extensions on a device gives out partners the edge they need to make these nuanced decisions.
What is the incremental value of using KYC automation for different business operations? Any specific case study that you would like to point out to our readers:
An automated, risk-based approach to becoming KYC compliant will always be able to save companies incremental amounts during the onboarding process, which can amount to huge savings over time. Depending on the IDV provider, the volume of users, and the depth of the checks being carried out, the average KYC check can cost anywhere from $1 to $5 per registration, including the ones that turn out not to be business-viable. Stacking IDV and KYC solutions with risk management software like SEON’s can save money throughout this process in two ways: cutting down massively on human resources allocated to the onboarding process and saving money on the KYC checks themselves through a system of dynamic friction. Having a comprehensive, automated, risk-based fraud detection system stacked into your onboarding software can make most customer journeys require little to no oversight, which also equates to a smoother customer experience than one that necessitates manual sign-off. This same software can also detect indicators that a new user will certainly turn out to be fraudulent and so can prevent that user from progressing deeper into your infrastructure before KYC checks, thus saving the fee that the check would have cost.
We recently shared a case study with great results Soft2Bet has got after using our software.
Soft2Bet is a company that offers iGaming solutions to online casinos, sportsbooks, and other gaming operators. Prior to partnering with SEON, Soft2Bet faced challenges with fraud prevention and identity verification, resulting in increased chargebacks and decreased customer satisfaction. SEON’s fraud prevention and detection tools enabled Soft2Bet to identify and block fraudulent activities, resulting in a significant reduction in chargebacks and improved customer trust. Soft2Bet was able to customize SEON’s user-friendly interface to their specific needs and maintain compliance with industry regulations and data privacy laws. This partnership highlights the importance of investing in robust fraud prevention and identity verification measures to ensure the safety and security of online gaming platforms.
ChatGPT conversations are everywhere. How do you see ChatGPT and other generative AI apps playing a larger role in your industry?
As I said, AI excels at crunching through vast amounts of data. SEON users already benefit from our Whitebox model, which improves the holistic fraud monitoring capabilities by picking out anomalies and suggesting rules based on malicious patterns it identifies. This is a more collaborative approach to integrating AI than simply having an AI assign fraud probability to every transaction without explanation. (We also have a blackbox model that can help with that, if someone prefers.) We ourselves are currently investigating integrating ChatGPT into our rule system to create super understandable conclusions, especially for the more in-depth checks we deliver, such as behavioral velocity checks. I think we will see increasingly advanced industry-specific generative models in the coming years, but they’ll work side-by-side with humans and support them in the tasks humans are not as good at. After all, looking through a database with 20.000 entries is daunting for a human. I don’t think machine learning models will take over end-to-end fraud prevention management any time soon, if ever. Of course, that may be more due to psychology than the actual capabilities of the tech.
AI and machine learning algorithms are helping financial services teams scale their user data management strategies in 2023. Could you highlight the role of AI ML and other emerging technologies in this specific domain?
All businesses should leverage ML solutions to get the most out of their valuable data. Across the millions of data points that any given user base might offer, there will inevitably be patterns that emerge. These patterns can be used, as with SEON, to fight fraud and assess risk. Still, they can also be applied to commercial optimizations like market segmentation and predicting future spending behavior. Many machine learning and AI tools are tuned to draw these valuable insights from data pools and allow finance teams to zoom in on specific pain points or opportunities to increase ROI.
Thank you, Tamas ! That was fun and we hope to see you back on globalfintechseries.com soon.
[To share your insights with us, please write to email@example.com]
At SEON, we strive to help online businesses reduce the costs, time, and challenges faced due to fraud. Whether you are a global financial leader or a small eCommerce, our solution simplifies fraud management so you can focus on what matters: growing and scaling your company.
Our talented team of consultants and developers is there to create a safer environment for online businesses. Cybersecurity is our passion. Anticipating risk vectors is our expertise. SEON is how we create a unified solution that combines ease of use, flexibility, and the ability to tackle complex problems in a simple way for your business.