Conscious of the ongoing arms race between worldwide money launderers and financial institutions, Guidehouse formed a partnership with Ayasdi in 2018 to deploy artificial intelligence and machine learning to more effectively detect financial crime. The collaboration combines Guidehouse’s compliance subject matter expertise and understanding of the financial sector’s regulatory environment with Ayasdi’s powerful machine intelligence platform.
Guidehouse and Ayasdi’s collaboration has thus far been successful. A large European financial institution engaged Guidehouse to review four years of correspondent banking activity for signs of irregular activity (a correspondent bank provides services on behalf of a domestic bank in a foreign country). Guidehouse and Ayasdi partnered to augment our traditional service offering with machine learning.
"Our Guidehouse and Ayasdi combined solution augments traditional, rules-based financial crime detection systems to help our clients adapt to everchanging threats."
While traditional detection systems segment customers based on predefined “types” (for instance, retail, corporate, or small business), Guidehouse and Ayasdi’s intelligent segmentation
used customers’ prior transaction activity to segment entities based on behavior. The machine learning model learned about customers’ behavior using transaction data and leveraged this data to identify new customer segments that better identify entities with similar transaction patterns. This allowed Guidehouse subject matter experts to better tune the system thresholds to identify anomalous and potentially suspicious behavior.
This behavior-based customer segmentation and segment-specific tuning reduced the alert population by 45%. At the same time, the machine-learning algorithm was able to generate alerts for each rule that were as much as 15% more productive — alerts that human investigators deemed worthy of follow up. The investigators worked more efficiently since they had fewer alerts to review, and the alerts were more likely to be associated with money laundering than the ones the system flagged without the benefit of intelligent segmentation. There was less time spent investigating false positives and more time spent on likely incidences of crime.
While such results should motivate financial institutions to adopt machine intelligence, many may not yet have the capabilities to deploy machine intelligence with ease and confidence. Successfully working with machine intelligence requires institutions to develop, document, and implement comprehensive testing and quality assurance protocols, and keep regulators abreast of its processes while following regulators’ guidance.
These tools are worth the investment in the fight against financial crime. Bad actors continuously learn and devise new tricks to launder money. Our Guidehouse and Ayasdi combined solution augments traditional, rules-based financial crime detection systems to help our clients adapt to everchanging threats. Indeed, our strategic partnership has already allowed us to help clients better detect and deter financial crime without compromising compliance.