Almost two years after Intel Corp. acquired Saffron, the chip maker is launching its first software product based on the cognitive computing startup’s technology.

Intel on Wednesday announced the launch of its Saffron antimoney-laundering adviser tool, which it claims is the first product on the market to use “associative memory” artificial intelligence in the financial services industry.

Money laundering is a big problem for banks and other financial services firms tasked with preventing it. The numbers say it all: According to the latest United Nations’ statistics, between $800 billion and $2 trillion in illicit cash – about 2 to 5 percent of the global gross domestic product – was laundered in various ways in 2016. The huge amounts of cash in circulation, and the data behind all of those transactions, make tracking it almost impossible, Intel said.

“The amount of data that banks and insurers collect is growing at massive scale, doubling every two years,” Gayle Sheppard, vice president and general manager of the Saffron AI Group at Intel, said in a statement. “While the quantity of data is growing, so are the types and sources of data, which means today that much of the data isn’t queried for insights, because it’s simply not accessible with traditional tools at scale.”

Of course, most banks still do their best to prevent money laundering where they can. The problem is that they’ve always been encumbered by older, limited software that can take weeks if not months to trawl through the sea of financial data.

That’s what Intel’s Saffron associative memory AI software is hoping to fix. According to the company, Saffron’s secret sauce is that it can mimic the way the human brain learns in order to create new associations and recall connected information. It connects information from a wide variety of data sources, surfacing anomalies and similarities that are normally incredibly difficult for human investigators to spot.

The Saffron AML Advisor uses its associative memory to detect money laundering by trawling through structured and unstructured data from emails, enterprise systems, the Internet and dozens of other sources. Moreover, the Saffron AML Advisor does all this in a transparent way, ensuring that financial services organizations using the software meet regulatory standards by providing logical explanations for the connections and recommendations it makes.

Intel said the software relies on a machine learning technique known as “continuous learning,” which means it doesn’t need to be trained by specific data sets first. You simply plug it in and let it go, with the software teaching itself on the fly, according to the company, and the method also helps ensure that insights can be surfaced faster.

Intel is kicking things off with the Intel Saffron Early Adopter Program for financial services firms that want to try out the software. The Bank of New Zealand has signed up as its first customer, Intel said.

Image: Intel