The COVID-19 pandemic greatly sped up TD Bank’s adoption of artificial intelligence, pushing the financial institution to utilize artificial intelligence in all lines of its business in 2020, according to Tomi Poutanen, the bank’s chief AI officer. 

“When the pandemic hit, we put our hands up and said, ‘How can we help?’ What the business needed was technology to identify which customers needed additional products, like mortgage deferrals, for example. That’s what we first jumped into,” Poutanen told The Logic on Wednesday. He is the co-founder of Layer 6, a startup that TD acquired in January 2018, reportedly for over $100 million, when it had just 16 employees. It has since grown exponentially, and now employs close to 60 people.

Over the past nine months, Poutanen said, TD has made significant advancements in its internal algorithms’ ability to predict a client’s cash flow and account balance, enabling the bank to better recognize when a customer might default on a pre-authorized payment for, say, a mortgage. 

“We designed a new capability in our digital app where we give a customer a nudge through a push notification if we know their balance is running low…. This is based on years of data that we accumulated looking at an individual’s spending pattern, income and upcoming payments,” Poutanen said.

TD’s residential-mortgage portfolio is a significant part of the bank’s business—it was worth $206 billion in its fourth fiscal quarter. In May, Evan Siddall, president and CEO of the Canada Mortgage and Housing Corporation, warned of a potential debt-deferral “cliff,” where those to whom banks granted mortgage deferrals might not be able to make their payments when the deferrals ended. 

Rizwan Khalfan, TD’s chief digital and payments officer, said the bank used Layer 6’s AI to “get ahead” of issues like mortgages. “The technology is not specific to whether a deferral runs out, but rather, it can predict ahead of time what account balances are going to be,” he said. 

TD has already started rolling out a nudge feature to select app users, not only informing them if they have a low balance or if bills are due, but suggesting how to invest their savings if the AI picks up on a consistently high account balance. TD claims its app is the most used retail banking app in Canada in terms of average monthly users, and saw the usage of the app increase 14 per cent year over year in 2020. 

While AI is used across all departments in the bank presently, Poutanen said the biggest opportunity the technology offers banks insight into customer behaviour. Features like the nudges result in a customer interacting with their bank more frequently, leading to greater retention rates—hence the bank’s recent focus on its digital app. 

“When you look at the core of what TD Bank does, it is a lender first. The decisions that AI can help propel are credit life-cycle decisions, and it requires really understanding the customer well,” Poutanen added. The bank only uses existing data within its system and data from certified credit bureaus to build AI tools that predict a customer’s needs—not their social media accounts. He declined to specifically quantify how much TD’s investment in Layer 6 has translated into monetary gain for the bank, saying only that Layer 6 has built 27 different use cases for TD since its acquisition. 

“I’m gratified that the engine we have built works great within TD. Our technology is used in the insurance department for identifying fraud … and in the American retail business, we have created a cheque-fraud model, where we look at cheque images to determine its authenticity,” Poutanen said. 

Many major financial institutions in Canada have made investments in artificial intelligence over the past few years—a direct result of Canada’s national AI strategy, Poutanen believes. RBC and its research lab Borealis AI recently launched an AI-based electronic trading platform called Aiden that uses deep learning to improve trading results. Last year, CIBC spent US$10 million investing in Charlotte, N.C.-based Stratifyd, an AI-powered data-analytics platform. Manulife Bank uses an AI fintech platform called Mai to track account balances and provide the insurer insight into customer spending habits. Scotiabank, too, has launched a new AI platform—Global AI—that gives customers personalized financial advice and helps the bank understand its own customers better.

“Any large enterprise not investing in AI is going to be left out,” Poutanen said, adding that he believes Canada, in particular, is uniquely positioned because of its constant supply of AI talent. 

“What I’m seeing now is every single bank building AI solutions, so I don’t think the large, bureaucratic governance structure of a bank are necessarily impediments. I also see the regulators being very inquisitive and supporting in this kind of work.”