TD Bank Group has acquired Toronto-based artificial intelligence startup Layer 6 as the bank continues to pursue new ways the tech could streamline internal processes and improve its customer experience.

Layer 6 focuses on using AI for prediction and personalization, primarily working in finance, e-commerce and media. Layer 6 founders Jordan Jacobs and Tomi Poutanen were also co-founders of the Vector Institute, an AI research centre launched last year associated with the University of Toronto and backed by funding from the Ontario government and more than 30 corporate partners, including TD.

The startup will continue to operate independently, but as part of the acquisition, TD will now be Layer 6’s sole customer. Rizwan Khalfan, chief digital and payments officer at TD, says the deal remains in its early days, but for the time being, the goal is to ensure Layer 6 can continue to work with “the agility and innovation” they already are.

“Our belief is we’re evolving from a mobile-first world to an AI-first world,” Khalfan says of the motivation behind the acquisition. “Every industry is transforming because of it, but banking in particular because there’s such large amounts of data.”

Khalfan said TD was drawn to Layer 6, in part, because it performed with 10% higher accuracy at last year’s international ACM RecSys Challenge. Also, the Layer 6 platform incorporates a great deal of transparency, explaining how it got to a prediction or decision, allowing humans to both rationalize the decision and put the result into action.

“When you talk about AI services, a lot of people describe it as a black box that spits out an answer,” Khalfan says. “But can you act on that answer? Who do you give credit to, or what’s the next best product for a customer? That’s incredibly important for us.”

There are four main areas Khalfan identified where Layer 6’s work will be incredibly helpful. They include simply processing the large amounts of customer data the bank has and streamlining its internal processes around things like product applications, both commonly pointed-to as rewarding use cases for AI.

But there are also two customer-facing applications that could also be put into place. One was using AI to customize and automate some of the billions of digital interactions TD’s customers have through its online banking platform, such as predicting when a bill payment is coming due and offering to make it automatically for a customer. The other, and one Khalfan says he is particularly excited about, is using AI to augment the expertise of its existing staff and advisers.

“There are so many advice situations in banking where this could be used,” he says. “When an investment line is coming together, an AI service could be assisting the adviser with prompts about questions to ask and suggestions about co-creating a plan for a customer based on their needs. Or it could help them better determine what the next best product for a customer could be. That’s the kind of thing where this would really come to life.”