The Toronto-based bank, which serves more than 17 million customers in over 30 countries, has 1,000-plus AI models for jobs like fraud detection, risk analysis and marketing. This new AI private cloud can analyze more than 13 million client records in under 20 minutes.
“We’re essentially applying better methods to an old problem to achieve a better accuracy,” said Foteini Agrafioti, chief science officer at RBC and head of Borealis AI, the bank’s research unit. “An increase of a couple of percentages in one of our businesses is very, very meaningful at the scale that we operate,” she added. The bank’s current library of AI models include many for core internal operations, such as models for setting credit limits for new borrowers, and models to detect signs of money-laundering.