In a context of digital transformation echoing with artificial intelligence, cybersecurity and customer experience, financial institutions are facing multiple challenges to adapt their services to new needs and expectations. Frank Roessig (Head Digital Solutions for Finance) and Claudia Albanese (Innovation Manager, Finance) from Telindus Luxembourg met with the BEAST team to discuss KYT – Know Your Transaction – which notably aims at preventing fraud through the use of AI.
Tackling the current challenges of KYC
« Telindus as a provider of a complete digital KYC – Know Your Customer – solution, is addressing some of the key challenges faced by the financial sector, from banks and funds to life insurance companies, but also other industries, like law firms, transport authorities, etc., and serving all types of clients, from startups via crowdfunding platforms to asset managers, » starts Frank Roessig. According to him, the current challenges of KYC are partly linked to a frustrating customer experience, which might lead to clients switching service providers. Also, recurrent tasks are costly for companies and are actually not bringing any added value to the customer. Finally, regulators all over the world are getting more stringent when it comes to applying the latest laws on money laundering and data protection. « In this respect, and in order to help our clients tackle these challenges, we have launched DigitalKYC, a new digital solution that enables the risk-based management of KYC, with several modules that can be plugged-in, thereby adapting DigitalKYC to all business models. It is able to cover individual KYC in over 130 countries and institutional KYC in over 80 countries, » highlights the Head Digital Solutions for Finance. Thanks to the Telindus-powered solution, clients now have a swift compliance journey on the one hand, while institutions are significantly cutting costs and enhancing the quality of the services on the other hand: « it’s all about creating the KYC of the future, and even the finance of the future! The latest development is addressing the increasing need to Know Your Transactions… »
Fraud detection or finding a needle in a haystack
« Nowadays, every asset manager, banker, or any other stakeholder, is experiencing an important digital revolution, » underlines Claudia Albanese. In 2018, only 5% of global payments were still done via paper: we are witnessing a change in the payment mix, with a strong movement towards non-cash based transactions. The expert explains it with the advent of online and mobile banks now offering swift and relevant customer experiences: clients can easily connect at any time and dispose of their financial services through an omni-channel accessibility. « Yet, in between a vast majority of transactions that are legit, you may find some fraudulent ones. In this respect, KYT consists in a constant monitoring of the transactions, in order to assess fraud, » she adds. On average, 99,85% of the credit card transactions processed every day are legit: therefore, detecting fraud is like finding a needle in a haystack. The Telindus experts decided to come up with a KYT concept leveraging the power of AI. Marketed as an « AI-powered tool box », it detects anomalies.
As explained by Claudia Albanese, « thanks to a powerful infrastructure, we are also able to run computational data to identify fraud within a network. The solution aims at responding to the AML and PSD2 regulations, among others ». She describes it as a natural evolution from KYC, leveraging on the data economy.
According to Frank Roessig, the future of finance is based on data and is now being extended through the AI-driven detection of anomalies in data. « These anomalies can be negative – money laundering, fraud, etc. – but they can also be positive and used to detect client needs and attrition, by monitoring their behavior and satisfaction. It is crucial to detect client needs and turn them into business opportunities. We all know that these tasks cannot be done manually, yet, from a regulatory perspective, financial institutions and companies now have to report suspicious activities, with a thorough and detailed description of each anomaly, » he adds. Frank Roessig also warns against «over detection», which could block the clients in their transactions and instantly affect their experience with an app, brand, etc. The experts at Telindus, when building such a concept, had to deal with a huge amount of data, and therefore needed significant computational power. « Then, in order to detect the 0,15% fraudulent transactions, we had to find the best AI-model which would give us the best prediction. The solution needs to detect in an optimal way, without blocking the users: clients most certainly won’t enjoy the fact that a transaction is flagged as fraud if it’s actually legit, » says Claudia Albanese. Her colleague then discussed the challenge of interpretation: anomalies can be detected, but people also need to be able to read them. According to Frank Roessig, this also includes a key « visualization » aspect, as compliance and marketing teams must be in a situation where they can easily understand what the data is telling them in order to take a sound decision. « It actually involves data scientists, IT managers, compliance experts, marketers… and we always add a design thinking component because the user should understand what’s behind the machine. The report must be decision-ready, » adds the Fintech expert. Moreover, since the solution builds on AI, it will keep on learning as time goes by and will eventually become better and more precise. As Frank Roessig puts it, « at the end, AI gives us recommendations. It is us, Humans, who are responsible for the decisions that are taken ».
The combination of tech and industry expertise
Currently, banks are using rule-based systems to detect fraud consisting in « if » and « then » statements. These can block multiple cases, but fraud, in its very definition, is constantly evolving. « Therefore, we should assume that we are not safe, and that fraud gets more sophisticated. As the CTO or CIO of a financial institution, I wouldn’t be comfortable knowing I’m only using a rule-based system that cannot account for all situations and does not evolve. I would feel safer with a system that monitors all transactions in near real-time, » comments Claudia Albanese. She also stresses that the Human expertise is still crucial since the solutions gives an indication with a confidence interval, with the user taking the final decision. The KYT solutions can therefore be defined as semi-supervised, keeping the people in the loop and building on their expertise and judgement.
To conclude, the experts discussed the benefits of opting for such KYT concept: it allows a constant monitoring of all transactions, done in a semi-automated way, with data being directly plugged into the system and scored with a confidence interval in near real-time. « A more secure way to deal with potential cases of fraud as it flags events that you do not expect to happen, » underlines Claudia Albanese. KYT surely maximizes the amount of anomalies detected: cheap to operate as you can start in an agile manner, it also allows the detection of new sources of revenue for the company. « Finally, the system gets better as it is used. We are witnessing the very same process within the Fintech team of Telindus, with a presence in Benelux and in North America, and an expertise and knowledge growing on a daily basis in the fields of data management, customer experience and cybersecurity, » concludes Frank Roessig.