How BMO uses AI to get more of business clients’ business

• Key insight: BMO is exploring a new use case for AI — using it to help salespeople have better informed conversations with middle-market business clients.
• Expert quote: “We like to be that customer’s trusted advisor, and this allows us to play that role in a more consistent way,” said Rose Grande, head of North American corporate card product and programs at Montreal-based BMO.
• Forward look: BMO is launching a pilot in Canada in a couple of months that should evolve into a rollout by the end of 2026.

BMO was looking to grow its mid-market business relationships, an effort that calls for a detailed understanding of a customer’s business.

Processing Content

Such insight is not always easy to come by, according to Rose Grande, head of North American corporate card product and programs at Montreal-based BMO.

“We’re able to do customer intelligence work ourselves on a very bespoke, one-off basis, but when it came to being able to scale it, we weren’t able to do so,” Grande told American Banker.

She and her team decided to try a technology from Codat that could pull the bank’s customers’ payments and accounts payable information through application programming interfaces and then use machine learning to generate recommendations for BMO’s salespeople. 

BMO’s collaboration with Codat comes at a time when most U.S. banks are considering AI use cases and taking AI pilots into production. In American Banker’s AI Talent Shift survey this year, 66% of bankers said AI is a strategic priority for the firm. 

“Banks surely want to use AI not simply to automate internal tasks, but to better understand client behavior, identify any unmet needs and make bank relationship managers more effective,” Bradley Leimer, founder and principal of Leimer One Advisors, told American Banker. “Commercial banking is a natural place for this because so much value sits inside cash flow, payments, working capital and the relationships of client suppliers. Much of that data has been fragmented across enterprise resource planning systems, accounting platforms, PDFs, spreadsheets, and bankers’ client meeting notes in CRMs. I’ve seen this for myself in some of the use cases I worked on at Sumitomo Mitsui Banking Corporation — the data is a mess to contend with, but the rewards are significant.”

What Codat’s technology does

Codat helps BMO obtain and analyze customer data it would have trouble getting on its own.

“Getting customers to share data is an innate pain point that we see all over banking,” Joey Rault, chief revenue officer at Codat, told American Banker. “Bankers, Treasury managers, Treasury consultants, payment consultants and sales specialists are always asking their clients for data to better serve them.”

Codat retrieves specifics about customers’ accounts payable, such as which suppliers they’re paying and how they’re paying them — “all of those types of insights that help us to advise our customers on best practices for payments,” Grande said.

Codat automates the ingestion of the data through the direct API connections it’s developed to draw information from more than 20 enterprise resource planning and accounting software programs, including QuickBooks, Oracle, NetSuite, Sage, Microsoft Dynamics, Workday and Xero. Using APIs to gather client data works far better than getting customers to send CSV or PDF files, Rault said. To gather data from accounting and ERP software for which it hasn’t built an API, Codat uses an “intelligent upload tool” that banks’ customers can use to upload raw files from those systems. 

The vendor uses machine learning to categorize the transaction files and generate personalized recommendations that are shared with BMO’s Treasury and payment sales teams. The salespeople use the data and recommendations to have better conversations with customers. 

“We like to be that customer’s trusted advisor, and this allows us to play that role in a more consistent way,” Grande said.

Codat can fill in certain blanks, Rault said. For instance, some ERPs don’t capture the payment method, e.g., check, ACH or card. Codat’s software looks at the invoices to try to find that information, and where it’s guessing or inferring, it says so.

“We use that information to say, hey, BMO, here are the gaps,” Rault said. “Here’s what we think is the payment method. That’s where we think AI is going — showing where the gaps are, and being clear on what we infer and how we got there. That helps the BMO team go have a more advisory conversation with the client: Hey, we think that you’re paying some of these vendors via check. Is that how you’re engaging with them today?”

About 15% to 25% of business payments are still made by check, Rault said.   

“Obviously there’s a lot of problems with checks,” he said. “One of the primary problems is speed of payment and controls on working capital, which is a huge focus for middle market businesses. A card gives you that much stronger control over the timing of the payment.”

There’s also a lot of check fraud, Rault noted, which could get worse as fraudsters use AI imagery to create fake checks. 

“That’s a huge opportunity for businesses and the bank to find check payments and move them to a more modern payment, whether that’s something even more antiquated, like ACH, or more modern like a card or even some of the more modern payment rails.”

As BMO looks to grow its share of wallet with its customers, playing an advisory role is important, Grande said. 

“So being able to get this information that’s been enriched and augmented by the AI intelligence, to go back to our customers and say, you’re spending X dollars on your corporate card program, we see that you could be spending this much more, and that would result in incremental benefits to you, whether it’s cash back, or rewards points, to really help them maximize that,” she said.

The Codat information also helps salespeople have conversations with clients about their money flow. 

“That’s really where our Treasury payments consultants add value to our customers, by saying, if you move this payment from cash or from check, you’re able to then save or build upon your working capital, and benefits through delayed payments and those types of things,” Grande said. “What we want to be able to do on a broader scale and in a very customized and repeatable way is to offer our Treasury consultants the ability to provide their customers the insights as to how they’re making payments today and how changing that could improve their cash flow.”

For instance, a customer may pay its suppliers in full within 10 or 30 days of receiving an invoice.

“That means they need to have the cash flow to be able to do that,” Grande said. “If they were to put that payment on a card that allows them to float that payment so the supplier would get paid immediately, but the customer itself would have to the end of their billing cycle to make that payment, that allows them to spread out their flow of dollars to be able to manage other more strategic business payments that they need to make.”

Early returns

BMO has deployed Codat’s technology in the U.S. and it’s launching a pilot in Canada in a couple of months that should evolve into a rollout by the end of 2026. 

“So we are fully in,” Grande said. “We’ve had customers who have used it and connected, and we’ve already started to see some success stories from customers.”

Since BMO started using Codat’s software, several clients have moved a chunk of their payments from check to card. 

“So we’ve increased their spend with BMO by about 45%, which is good for us, but it’s also good for the customer, because the customer will then receive the benefits that the BMO program offers, but also has not had to significantly extend themselves from a cash flow perspective, they’ve been able to save on that,” Grande said. 

Leimer expects more banks will adopt this approach. 

“The broader opportunity is not just AI-generated recommendations, but relationship-driven intelligence using permissioned data to understand where the customer may have friction, whether that is trapped in working capital, payment risk, fraud exposure or greater operational inefficiency,” he said. “This can improve client relationships and drive more bank revenue by making client conversations more relevant and contextual. Instead of generic B2B campaigns or periodic banker check-ins, the bank can come to the client with something specific they see in their payment flows, and suggest ways they can help. And leveraging APIs to draw in this information helps provide that timely check-in that, when infused with AI prompting, can really make a bank relationship stand out.”