Banks have always been in the tech business, and they’ve always had to do so with a higher-than-normal degree of care. The Silicon Valley ethos of “move fast and break things” doesn’t really apply when millions of people’s financial futures are on the line—especially when it comes to technology as transformative as AI. 

The rapid acceleration of artificial intelligence holds enticing—even exhilarating—promise for the financial services sector. From an operational point of view, AI can improve a bank’s accuracy, accelerate its decision-making, and enhance its efficiency; looking through a customer service lens, it can create faster, safer, and more convenient ways for people to manage their money. At root, AI offers banks the opportunity to do what they do with a lot less friction—and with a lot more upside. McKinsey estimates the potential value of AI to the global banking industry at US$1 trillion; the productivity gains caused by generative AI alone could unlock hundreds of billions of dollars. Perhaps unsurprisingly, the financial sector invested US$35 billion globally in AI last year—a number set to nearly triple by 2027.

Yet the robot revolution won’t conquer banking overnight. Canadian banks have still-nascent ethical guardrails to consider, along with evolving risk-management provisions. In practical terms, scaling advanced technology across large, complicated organizations, with multiple lines of business and tens of thousands of employees, involves significant, and material, logistical hurdles, exacerbated by an industry-wide talent crunch for key technical roles.  Furthermore, many consumers—whose trust is the bedrock of any bank—are still pretty skittish about AI.  

Given the pace of change and the stakes at play, Canadian banks find themselves at a critical nexus: they must adapt, but they must not do so recklessly. While there is no playbook, there is much to learn from those who have experienced some early wins. Take BMO Financial Group: The 207-year-old bank has developed a thoughtful AI innovation strategy that balances progress with prudence—and that is positioning the bank to strategically reap the benefits of this bold new era.

A history of applied innovation

You might not automatically associate the steady-Eddie realm of banking with the cutting edge of technology. But think, for a moment, of the sophisticated technological backbone required to, say, support the launch of the first credit cards 60-odd years ago, or power up the early electronic exchanges of the dot-com era, or to develop to the smartphone apps that allow you to send money to a friend with a few taps of your thumb. 

BMO is approaching AI by applying the same ethos that led it to launch North America’s first “computer-based, fingertip banking system” 55 years ago: technology should only be deployed if it’s in the service of clear business and organizational goals. “Our business is to be a financial services company, which, to me, has two parts: to serve customers and to manage risks,” says Raphael Schapiro, COO of BMO’s Technology and Operations Group. “Our job is not to invent things or to do innovation. Our job is to apply innovation that advances our strategy.”

The bank assesses whether to use AI in a tool or process by evaluating the initiative against a set of mandatory criteria. It must have a defined business use case, in the form of generating revenue or improving productivity; it can’t just be something interesting or fun to do. It must have rock-solid technological underpinnings that work within the bank’s ecosystem; it can’t just be the domain of niche specialists. And it must hold up against a litany of internal stress tests related to cybersecurity, risk, privacy, and legal, among many others.

If an off-the-shelf solution exists that passes all these tests, the bank will adopt and/or adapt it. If not? Then, and only then, will the BMO team build it themselves. And they almost never start from scratch.

A focus on practical progress

The AI development and adoption process at BMO is, like many organizations, a group effort. The bank collaborates on research with the University of Waterloo, Queen’s University, and the Vector Institute to complement a strong core of internal talent that explores and evaluates AI developments. They are, as you’d expect from folks at the forefront of technological transformation, very interested in breaking boundaries in new and, ideally, patentable ways—but never for the sake of doing so.

“We never start thinking that what we’re doing is a patentable solution,” says Lawrence Wan, BMO’s Chief Architect & Innovation Officer. “We build and implement things that the business needs, and only once we’ve actually put it into production will we reflect on our work and say ‘Hey, wait a minute, this is a really creative way of doing things.’” 

Take a recent initiative to improve cheque-scanning capabilities, for example. The bank trained an AI model to zoom in on the key sections of the image, bringing together combinations of models to locate the signature, decipher the signature, recognize a diverse range of names you might expect within Canada, and map it to a customer name. In isolation, these functionalities had all been done before. But the specific combination proved to be something fresh, and—as it happened—patentable. “That’s how it works with the majority of our patents,” Wan explains. “It’s about the combination of things in totality, and how we apply them.”

Until relatively recently, banks might not have bothered trying to patent advancements such as these. In fact, financial institutions have traditionally kept pretty quiet about their technological breakthroughs. “Banks tend to be very conservative and very numbers-driven, so their old strategy was to be in stealth mode,” explains Isi Caulder, a Toronto-based patent lawyer and Partner at Bereskin & Parr, where she co-leads the firm’s AI and cleantech practices. Tried-and-true means of protecting intellectual property, like quietly maintaining trade secrets, would often suffice. But things have changed in the past decade, as the strategic value of AI to the finance sector has intensified, and as patent law has evolved to increasingly recognize intangible computational innovations. “IP is really important right now,” Caulder explains. “Fences are going up all over the place.” In relatively short order, a strong AI patent portfolio has become a competitive advantage. But “strong” doesn’t necessarily mean “huge.” 

A platform for future success 

BMO has applied for more than 70 patents since 2018, and earned 22. Some relate to AI features that improve computer functionality (e.g. a novel way of connecting the bank’s IT infrastructure); others are the result of a clear practical application (e.g. a new probabilistic algorithm that enables immediate visibility into a client’s activity across multiple lines of business). Other banks have more, but BMO’s deliberate quality-over-quantity philosophy allows the bank to leverage the benefits of patents while keeping its eye on the prize.

“BMO is punching above its weight when it comes to patents,” explains Alexandra Mousavizadeh, Co-founder and Co-CEO of Evident Insights, the U.K.-based research firm behind the Evident AI Index, which benchmarks the AI capabilities and maturity of 50 major global banks. BMO placed 16th overall on the most recent index, and ninth in the innovation category, which assesses patents, research, and related variables. Moreover, the bank’s patents boast more than twice the average number of forward citations of other evaluated institutions. That means BMO’s patents have influence, Mousavizadeh says; furthermore, it signals a holistic approach to innovation within the bank. “The ability to generate highly cited patents means that you’ve got really good talent that can sit and think through something really novel that is important for the financial services sector,” she says. “It points to a number of strengths in the bank.”

Which is, in many ways, the point. Aside from the most obvious benefit—to protect hard-earned intellectual property that holds competitive value—BMO’s patent portfolio serves as a sort of stake in the ground. “A patent is a metric of innovation,” says Eric Morrow, Managing Director of the Enterprise Data Science & AI group within BMO’s Data and Analytics arm. “It helps us to articulate the culture of innovation we have within the organization to external parties, whether that’s our customer base or our shareholders. And it signals to our employees that they, too, might have an opportunity to earn a patent; it creates a strong incentive for thinking outside the box.” And a little can go a long way.

This considered approach helps ensure that the bank is applying AI in a thoughtful, strategic, and effective manner today, and that it won’t overshoot its skis as the technology continues to evolve. “If we were to be a patent factory, that would serve no one’s interest,” Morrow explains. “You don’t do AI to do AI. You do AI to support a business strategy.” 

This content was paid for and directed by BMO Financial Group and was produced independently of The Logic’s newsroom in consultation with the advertiser. You can read our policies on advertising, sponsorships and partnerships here.