Silent Eight Preps Transaction Monitoring Tool

Regtech vendor Silent Eight is preparing a transaction monitoring tool that will be rolled out in Q4 2020.

Martin Markiewicz, CEO at Silent Eight, says the tool, an extension of the company’s existing Name and Transaction Screening Optimization service, is already being used by some of its customers. 

“The only reason we’re not scaling this use-case yet is that I want to make sure it’s a nicely-packaged use-case—almost like a box into which you can just plug in your cables and start using it,” Markiewicz says. 

The tool is about 90% finished, he says. Currently a select group of users is testing it to make sure that when they do roll it out, “if we want to deploy it to 100 different places at the same time, there will be no issue with that,” he says. 

Through its offices in Singapore, New York, London, and Poland, Silent Eight helps tier-1 and tier-2 banks with name, entity, and transaction alerts, as well as investigations across various business lines in the bank, from the retail side to wholesale and investment banking. 

In 2018, Standard Chartered became Silent Eight’s first client to deploy the screening solution globally. In November last year, Standard Chartered—through SC Ventures, its fintech investment unit—participated in Silent Eight’s Series-A funding round, in which it raised $6.2 million. 

Artificial Analyst

Silent Eight was formed in 2013. Today, it has a staff of about 80 employees, 90% of which are data scientists and software developers. 

Its NTSO offering uses artificial intelligence to speed up the time it takes to conduct investigations into suspicious events, and flag those that will need a human-led, in-depth investigation. The engine can process 2 million alerts per day, and has been approved by regulators in over 60 jurisdictions.

Markiewicz says the AI model is actually a composite of several different models, including deep learning, but the main AI driving the tool is “a twist on the concept of decision trees.”

“The idea here is that the final decision needs to be done in a transparent way so that you can actually understand how the final logic and the final argument was derived. So that’s why it couldn’t be done just by applying a massive neural network,” he says.