Napier AI has launched a new artificial intelligence capability aimed at improving anti-money laundering investigations.

The feature, known as Insights AI, forms part of the company’s transaction monitoring platform and is designed to provide analysts with clearer explanations of customer behaviour linked to suspicious activity alerts. The tool surfaces AI-generated insights directly within monitoring tasks in order to highlight behavioural patterns and potential emerging risks during an investigation.

The development follows testing carried out under a project known as Theseus, which was presented at the Financial Conduct Authority’s Supercharged Sandbox showcase. The project explored the use of frequency-based algorithms applied to large-scale synthetic financial datasets to identify potential money laundering typologies.

The resulting models have been incorporated into Napier AI’s Continuum platform and underpin the new Insights AI capability.

Dr Janet Bastiman, chief data scientist at Napier AI, said: “Participating in the FCA Supercharged Sandbox allowed us to design and run new approaches to testing AI models for anti-money laundering. One of the biggest historical challenges in tackling complex money laundering typologies is the disconnected nature of the data required for pattern analysis along the complete lifecycle of a customer behaviour or transaction flow.”

“I was inspired by nature to find the flow, by learning from my peers in climate research and applying the fluid dynamics of sampling river water to the way money moves through the financial services ecosystem. We developed a methodology to detect the ripples in the transaction flow where money had been injected, even if we're sampling downstream of that. And by applying the concept of the frequency and amplitude of waves to these “ripples” we were able to identify financial crime even when we were sampling further ‘downstream’ from the placement of illicit funds.”

Napier AI has previously worked with the Alan Turing Institute and Financial Conduct Authority on a synthetic data initiative designed to support the development and testing of financial crime detection models.

---