Despite billions invested in legacy AML solutions to stop financial crime, AML systems still fail 80% of the time. Quantiply provides a suite of automated AI applications so banks can fight financial crime, be more efficient, mitigate risk against damage to reputation, client trust, market share, and reduce false positives by more than 50%
As an ever increasing number of fintech companies make an already competitive market even more so, banks are being forced to look for ways to improve the effectiveness and efficiency of their business. AI helps banks improve their bottom line with the people they already have and the data they’re already collecting.
Quantiply Sensemaker is designed to help financial institutions understand their customers at a significantly deeper level, recognize individual risk patterns in real time and efficiently filter good customers and their transactions from bad actors and illicit activities. Quantiply’s innovative Explainable AI simplifies compliance requirements and delivers a number of benefits to banks, as well as to the regulatory agencies themselves.
Today, bank AML departments rely on two types of analytical solutions to comply with AML and KYC requirements and manage risk within their institutions.
These include solutions such as KYC screening tools and AML transaction-monitoring systems. All of these solutions require numerous AML analysts, and in many cases, the support of external specialists to create rules that identify and trigger alerts for suspicious activities and transactions. Unfortunately, keeping the rules up to date can be a time-consuming challenge that may take three to six months to complete, and yet still results in an unacceptably high rate of false positives. In addition, the rules that are created apply only to what’s already known, so they can’t detect emerging patterns, insights or anomalies that lead to added risk exposure for a bank. Another problem is that the rules-based systems that trigger alerts remain inadequate, requiring two to three hours to investigate each case and fewer than 15 percent complete customer due diligence in under one hour.6
These solutions focus on using machine learning to discover anomalies or transactions that stand out from the norm, which the human eye might not be able to detect as easily or quickly. This new class of solutions based on machine learning holds promise, but still falls short in critical areas, as many banks can attest after conducting their own experiments. While they can more easily detect new attacks, they still generate far too many false positives. In addition, training the systems, labeling the events and getting the systems to run at scale in production with the desired uptime requires specialized skills provided by highly trained (and highly paid) personnel. In addition, the typical approach for this type of machine-learning-based solution renders decisions in a ‘black box’ fashion with no visibility into a system’s internal workings. This simply isn’t acceptable to regulators, who require decisions driven by machine-learning-based systems to be easily explainable—ultimately leading to a need to invest in more resources with costly, specialized machine learning skills.