Acquiring an Edge: TD Innovation Surges Ahead with AI

Interacting with your finances can be described in two ways: Incredibly interesting, or unequivocally boring. Shopping on Instagram and stock trading on the subway is a stark contrast to the teller lineup on a busy lunch break.  In the age of tap-to-pay, digital wallets, and one-click shopping, it’s easy to lose track of every dollar leaving your wallet, let alone which companies have access to it. There are countless ways to spend and manage finances, and as more fintechs and payment solutions enter the market, navigating the intersection of technology and money will only get more confusing.

Payments companies like Mastercards and Paypal are transforming the way we spend our money, but those systems are useless without a bank at the foundation. Banks leverage every partner and piece of technology they can to drive personalization, convenience, and security.

The most engaging ideas surrounding big banking are not actually the specific technologies or features they have on offer, but the methods as to how they onboard cutting-edge innovation. Every bank has a chatbot now—in 2019 it is all but a standard feature. But these massive corporations rely so heavily on general trust and mutually assured confidence from their customers that any kind of “innovation” is usually stalemated by a slow rate of adoption and compliance—the first chatbots were around 15 years ago, and are only now just rolling out to banks.

The seemingly slow pace is not due to lack of trying; every major bank is investing in innovation at some level. The TD Bank Group has well over $1 trillion worth of controlled assets and employs over 81,000 globally making it the sixth-largest in North America. With so much under control, TD must be extremely careful when rolling out new technologies. Even still, what does ‘careful’ mean when thousands of employees get their hands on a new feature before it even sees a single customer interaction?

Careful means a lot to Michael Rhodes. He is TD’s group head of innovation and technology, and below him is a network of change, including TD’s Innovation Centre of Excellence, two separate tech labs in Waterloo, a design lab in Toronto, and a cybersecurity lab in Israel.

“What the banking industry is grappling with now is how to graph a legacy of trust onto a new banking backbone powered by technology and customer data,” Rhodes wrote in Techvibes late last year. “It’s no small task…But we see a way forward that not only faces these challenges head-on, but that can power the change that’s needed today and into the future.”

“We’re replacing a lot of analytical systems with deep learning systems that are shown to be much more effective” – Tomi Poutanen, co-founder and CEO of Layer 6, acquired by TD

One member of Rhodes’ team is Tim Hogarth, the VP of TD’s innovation framework. Part of his role is to create a guide for how the bank approaches partnerships and innovation opportunities outside the bank’s walls. TD—particularly through its Ventures arm—has reached out to over 1,500 different fintechs in hopes of finding the right fit to build value for customers.

“What we do is we look at a business problem we’ve got,” says Hogarth. “We start with that problem and go looking around the world, and it’s quite a different response based on where we’re looking. With small companies, it’s important to be nimble, it’s important to be pragmatic, and it’s important to operate well within the bank’s risk appetite. But we have different processes to how we start those partnership conversations.”

For example, in September 2018, TD announced a licensing agreement with Hydrogen, a fintech platform designer. TD will use the platform to provide its customers with an end-to-end digital investing experience.

But to get to that partnership, it took due diligence. TD looked at over 30 different potential companies to work with before selecting Hydrogen. TD Ventures will engage the first round of conversation and evaluate a company quickly, then often make a gesture, whether it is a small equity investment, or donated legal fees to cover a patent through TD’s Patents for Startups program. Regardless, it’s integral that decisions are made quickly on both ends of the table.

“We facilitate these meetings with partners and we try to get people inside of the bank to a quick yes or a quick no,” says Hogarth.

Looking inwards

With a worldwide workforce of 81,000-plus workers and a healthy labs ecosystem, one might ask why a company like TD is even trying to engage with outside partners. Why not fix problems with the 1,000-plus engineers and developers already on staff?

“It really depends on the nature of the problem,” Hogarth explains. “We have a large and confident tech team, but some of the teams also have a very specific skill set that lends themselves to certain types of innovation. For example, our Waterloo labs focus on the mobile experience and usability. For other difficulties that look at different go-to-market problems, sometimes it involves finding a more complete solution, and often those are better served by small companies that are outside the bank and are thinking up novel ways of reaching our customers.”

This is where the beauty of innovation frameworks for enterprise companies come in. Innovation is not static—it demands to be considered from every angle, and with every potential partner possible. Taking too long to develop an in-house solution could spell disaster if a competitor rolls out a feature first. Similarly, opening the doors to widespread partnerships means inviting can introduce new risk.

“It’s about having the best talent internally and cultivating an innovation mindset, but also having the recognition that we should and will look outside our doors and borders to other places where there are pockets of talent that we can leverage,” Hogarth explains. “A good example is in cybersecurity. We have a number of in-house experts, but our centre in Israel, it’s in a vibrant hub of experts. That’s a great ecosystem to be close to, and to acquire technology there is a much better use of capital than developing it ourselves.”

Of course, in-house developed proprietary solutions to problems exist as well. Operations like TD Labs operate as a resource anyone in the bank can access. Business teams, typically led by executive sponsors, will identify a problem they want to explore, then bring it to the Labs division. If it’s an intriguing enough problem to tackle, the Labs will run with it for 10 to 12 weeks. The objective here is not to create a scalable market solution, but rather to prove a point.

“The main purpose for Labs is to explore ideas,” says Hogarth. “It’s not judged by how many things go to market, but how many ideas are successfully explored. If you try and take very small and uncertain problems and throw them into a large bank, it’s very difficult.”

One of the key takeaways for Hogarth and the innovation team is not based upon fully-realized business solutions, but rather team engagement. A key mandate for TD is to enable as many teams as possible to run with their own business ideas and see what can be developed within the bank’s own framework. This kind of end-goal ensures that the bank’s robust Labs and tech teams are always looking to improve their existing offerings, even as TD looks outside its walls for new partnerships.

The future department

Another method when it comes to building an innovation framework is a pretty simple one. Buy it. And that’s what TD did in early 2018 when they acquired the 17-employee AI firm Layer 6 at a $100 million price tag.

That might seem like a lot for such a small company, that demonstrates the value of meaningful innovation at TD. The bank was at the perfect intersection to make a deal: they existed as an organization with huge troves of data and an incomplete data science department. Instead of hiring AI capability, they bought it. With that talent comes deep roots in the AI community as a whole. Tomi Poutanen, co-founder and CEO of Layer 6, is also a co-founder of the Vector Institute, a world-renowned AI research organization home to the godfather of deep learning, Geoffrey Hinton.

On Layer 6’s side of things, it was as close to a no-brainer as it can get. Layer 6 is not acting as consultants or an advisory firm—they maintain their roles as engineers, working on complex data problems every day. The acquisition was obviously an incentive, but as an already-successful AI firm, Poutanen and his team joined TD because they had what every AI company needs.

“Deep learning is driven forward by access to massive unique data sets,” he says. ”You have to ask, ‘What are the unique data sets TD has access to?’ First of all, the sheer size of their retail banking base—25 million customers with a strong concentration in Canada and the U.S.—with capital markets, TD Securities, Waterhouse and Ameritrade, we have access to a huge book of assets under management.”

And, as Benji Sucher, the COO of Layer 6 describes it, the startup “went from a company that would look to scale a handful of different products, to a company with one customer with many different opportunities and engagements that we can get very deep with.”

TD also boasts mobile apps with over eight million active users, providing an opportunity for Layer 6’s researchers to develop algorithms that affect real users. Layer 6’s AI systems are deployed at the heart of TD’s data lake—every piece of information the bank is accessible to the AI team and allows them to drive personalization and recommendations.

“It’s about having the best talent internally and cultivating an innovation mindset, but also having the recognition that we should and will look outside our doors and borders to other places where there are pockets of talent that we can leverage.” – Tim Hogarth, VP of TD  Innovation Framework.

Layer 6 is currently operating as an independent company within TD, working out of the MaRS Discovery District, one of the largest innovation hubs in North America, and also the home of the Vector Institute. This separation gives the company some autonomy to pursue research endeavors, which they have succeeded at—Layer 6 won the prestigious RecSys challenge back-to-back, the first ever company to do so.

After the acquisition, a lot changed for Layer 6. Instead of having to focus on selling products to clients and maintaining the typical startup hustle, Poutanen could translate that hustle to research and outreach and funnel it back into TD’s innovation framework. With only one client to focus on, Layer 6 switched their approach and discovered new ways to understand TD customers.

“We have a tremendous amount of data on our customers, and historically, I would say our ability to turn that data into insights for customers has been limited by algorithms,” says Poutanen. “We’re replacing a lot of analytical systems with deep learning systems that are shown to be much more effective. What’s making them effective is that our new frameworks can take in more data. That is driving more accuracy.”

If there was a light at the end of the tunnel for how Poutanen could apply AI to TD’s innovation framework, it revolves around making every TD customer more confident when it comes to handling money.

“A big part of the goal is financial health,” he says. “The bank deals with a lot of people who have low financial literacy. That’s something that can be solved with data and learning from customers who manage their finances well. We can enable the broader population to learn from that. It’s gratifying to work on enabling customers to meet their financial goals using AI and a digital app that is highly personalized. It takes the conversations between a bank and the customer to a new level.”

Even though Layer 6 does not operate as AI consultants, TD still leverages the firm to improve how other companies factor into its framework. The bank works with hundreds of different technology partners and is in discussion with thousands more at any given time, so it would be wasteful to not utilize the top AI talent within the bank to vet some of the potential partners.

“We’ve identified opportunities for what we would call missing data or missing information that TD in of itself does not collect or generate,” says Sucher. “We identify it, and then see who can offer the information or expertise that would complement our product offering, so that together we could provide a unique personalized experience.”

Through all of the different ways to build an innovation framework, to Poutanen, it all comes down to one thing: bringing the best talent possible to work for Layer 6 and TD.

“It’s about attracting the best talent, as there is a limited set of people who can do this deep learning,” he says. “We are also carefully curating the type of work that has great impact so we can build a great environment and support the right academic feeder system.”

And what better way to make a great impact than to work in healthcare? In late 2018, TD and Layer 6 teamed up to support a University of Toronto initiative to use machine learning to better treat diabetes. This is a core example of the Layer 6 teams exploring outside research opportunities, something that would have proven difficult prior to the acquisition.

These kinds of initiatives are what prove TD’s innovation framework as one of the strongest in modern banking: Identify key investments outside of the bank, allow them to engage with opportunities that both enable social wellness and cutting-edge research, and use those examples to drive talent and innovation back to the bank. Let the smartest people in the building do the work they want to do because as long as it is innovative, it builds out the framework and positions the bank as a leader in the technology world.

“As a founder, before being acquired, we were mostly selling to banks,” says Poutanen. “Anyone who has sold to banks knows it can be a painful process.”

“Now I don’t have to sell. I can build, do the science, grow the team, and get to work.”