Banks Look to Get Out Front on AI
Testing the Waters
It’s called Liberty Live.
Dave Glidden, president and CEO of Middletown, Conn.-based Liberty Bank, describes this as a twice-a-year, in-house ‘broadcast’ where team members are updated on what’s happening at the institution, complete with a ‘fireside chat’ element during which he answers questions from employees.
It was during a recent episode of Liberty Live that Glidden addressed the subject of AI in the financial services sector and started by saying, “it’s no longer something that’s coming … it’s here.”
And the fact that it’s here is one of the motivations behind the bank’s creation of what it calls its AI Center for Excellence, a dedicated function designed to bring advanced artificial intelligence to every department of the bank.
“What I’m looking to do with AI is see where we can apply it in use cases to increase our efficiencies,” Glidden said. “And we’re going to follow what I call ‘ethical AI,’ meaning there will always be a human in the loop — there will always be a human making the final decision.”
Dave Glidden recently told his team at Liberty Bank that AI is “no longer something that’s coming … it’s here.”
At present, the bank is using various AI products, with names like Copilot, Salesforce, ABBYY Value, Rabbitt, and Lama, to create efficiencies and save team members time in areas such as document processing, construction loan automation (the bank has a strong niche in the timeshare construction realm), and smaller (for this bank) commercial real estate loans.
Not all banks have a formal center for excellence when it comes to AI, but they are essentially doing the same thing as Liberty — looking at opportunities to use this emerging technology to improve customer service while also shaving hours off the time it takes to do certain things, thus giving employees opportunities to make better use of their time.
And while doing that, they’re also careful to build thick walls between the AI products they’re using and the vast stores of information on customers and employees alike.
The opportunities, risks, and, yes, controversy concerning AI were addressed in a summary statement on the subject from Kathleen Murphy, president and CEO of the Massachusetts Bankers Assoc.
“Artificial intelligence offers significant opportunities to enhance efficiency and the customer experience across the banking industry,” she said.
“As banks explore these technologies, they will continue to apply the same disciplined approach to risk management, oversight, and due diligence that has guided the adoption of new banking technologies for decades.
“While AI can help banks serve customers more effectively, it cannot replace the human judgment, expertise, and relationships that remain fundamental to banking,” she went on. “Maintaining customer trust will continue to be the industry’s highest priority.”
Chuck Leach, president of Lee Bank, concurred with that sentiment, but added quickly that AI is not like the new banking technologies that have emerged over the decades.
“AI is different — it’s a quantum leap forward from other technology advancements that we’ve all experienced in our lifetime,” he told BusinessWest. “I guess all those other technologies might have made employees better or more effective, but this is just a quantum leap forward in terms of amplifying skill sets and enabling bankers to focus on the human interaction because some of the other tasks that are little more menial will be made much more efficient.”
Indeed, this is a rapidly, as in rapidly, changing landscape, one where best practices are a moving target at best, with banks and credit unions creating their own rather than trying to follow others, and return on investment is still difficult to calculate and often measured in hours or minutes of time saved rather than dollars, although those are added up as well, said Drew Weibel, Lee Bank’s chief information technology officer.
“We’re looking at hundreds of hours, easily, saved thus far,” he said. “It’s in small bits and pieces, but it adds up across the board.”
For this issue’s focus on banking and financial services, BusinessWest takes a look at how some institutions are deploying AI and how the technology is changing the industry, and in the course of doing so, we’ll look at the many aspects of rolling out this technology, from determining what products to invest in to how to measure results.
Technically Speaking
As he talked with BusinessWest about the AI Center for Excellence, Glidden described it as an effort to get out front on this technology, rather than be a ‘strategic follower,’ as he originally planned.
He described the initiative, undertaken in strategic partnership with Flare AI, as a hub for AI strategy, governance, and execution, with each of those elements carrying significant and equal weight.
“We have to make sure that we, as a company, from a governance standpoint, are controlling this,” he explained. “We need to have the policy rules in place but, more importantly, the right people around the table that can evaluate these things, whether it’s part of vendor analysis or identifying opportunities to improve efficiency with AI.”
The center will be led by Jeremy Miller, Liberty’s chief operating officer; Paul Young, chief financial officer; and Troy Damboise, chief enterprise risk officer, and co-chaired by David Hadd, Liberty’s head of Business Transformation; and Mike Stevens, senior vice president of Data Management and Enterprise Architecture.
The center is exploring ways to put AI to use to create efficiences with both back-office functions and customer-facing operations, Glidden explained.
“As much advancement as we’ve made here in my tenure, we still have a lot of manual processes and laborious ways of doing things, policies and procedures, and all the rest,” he noted. “So I really see AI for us, in these early stages, as bringing it in and learning how to apply it safely, from an informational protection standpoint, but also taking advantage of huge opportunities to improve our efficiency.
“If you can free up teammates’ time for more valuable tasks, then we can continue to grow and not have to continue throwing headcount at it,” he went on. “I don’t see it from a headcount reduction standpoint; I see it as increased efficiency for our teammates.”
And some current uses are creating such efficiencies, he noted citing Rabbitt and another product called Lama, a model that focuses on smaller commercial real estate transactions.
With the latter, the AI technology is helping to underwrite and monitor loans more quickly and more efficiently, Glidden explained. “We lend in 36 different states across the country, so it’s a big operation and a great business for us. But anytime you’re making loans that large for complicated projects, there’s a lot that goes into monitoring and underwriting them.
“This is an AI module that actually does all that monitoring for you,” he went on. “Instead of me having to put six teammates who are buried in the data of construction advances, progress payments, and all the rest, now they’re freed up because AI does it automatically. What would normally take a teammate maybe six hours … they get it in a better format in about 30 minutes.”
As for Lama, it addresses a gap in what Glidden called the “commercial real estate investor side.”
The bank handles many large transactions, but in Western Mass. and Connecticut, many borrowers are targeting smaller real estate deals, he explained. “It’s hard, when you build a machine for big, big deals, to put a small one through, and vice versa, so this is an AI we brought in that does the underwriting for small-investor commercial real estate.
“It’s not taking the time of credit analysts, portfolio managers, relationship managers, and underwriters that are working on a $25 million or $50 million deal,” he went on. “This automates it — once you get the information from the customer, the AI model reads it, does an underwriting of it, and spits it all out. In a matter of 15 to 30 minutes, there’s a very well-underwritten document that a credit person and a real estate person can make a decision on.”
Change Agents
Leach bristles at the notion that banks must have size to be efficient, a common refrain among institutions in pursuit of scale.
“I consider myself a really innovative CEO and president — I followed technology closely when I was an investment analyst, so it’s always been near and dear to my heart,” he told BusinessWest. “Stack on to that the knock on small banks, that they’re inefficient, that we need to have scale to be more efficient — and I always push back against that — and in the middle of this comes the next phase of AI, and we’ve really just embraced this whole concept of how to apply it at a community bank.”
Leach said the bank is being entrepreneurial, but also spirited, in its approach to AI, with creation of a committee to review opportunities and products, and a competition within the institution to identify new and creative uses for Microsoft Copilot — now being used by most banks to accelerate routine work — and other platforms.
“We’ve game-ified the use of AI and created a use case contest, for lack of a better word,” he noted. “People submit their use cases, tally up the hours, how much time or money they think it’s saved, and we have our AI, or IT steering committee, vet those, and there are prizes for those who have submitted the best use case.
“That’s a good starting point because, out of that, across the organization, we’re seeing some very unique and exciting use cases,” he went on. “Whether it’s in our commercial lending area or collections area or with our foundation and sorting through grant requests — we’re not dehumanizing it, but making it more efficient.”
Meanwhile, the bank has created a dashboard outside this game-ification of AI to tally up all the use cases and the hours saved, he went on, adding, “it’s almost like a national debt clock tracking the ROI with this whole endeavor.”
As for those use cases, they cover several areas of bank operations, from vendor management to the use of agents, such as on the retail side of the operation.
“Our branch staff have an agent that they can pose questions to. Those questions used to mean interrupting the branch manager or calling the operations area and talking to someone there,” he explained. “This agent can guide the human through those steps.”
Leach cited another use case, this one involving annual reviews on company compiled by the chief credit officer.
“It’s like writing up a Harvard Business School case study on a company — there’s a lot of detail, a lot of which can be carried over from the prior year’s writeup, but then, you have to integrate it with whatever new has happened,” he explained. “It’s a tedious process, and there’s a lot of hunting and pecking and going back and forth between documents. There’s a way with AI to speed up that process, helping with writing the narrative, bringing new data in but leaving the correct data from past years in there … it’s a way to craft a new credit memo or annual review on our commercial credits that’s cutting the time down from five hours to one hour.
“AI is doing the heavy lifting around non-knowledge worker stuff, such as transferring data and things like that,” he went on, “while the human, the skilled credit professional, is reviewing that and shaping it for the final product.”
There are many other examples of how is AI is shaving hours off processes and creating time for team members to put their time and energy to better uses, he continued, adding that, at this still-early stage in the work to deploy AI, return on investment is mostly being measured in time, not dollars, although it can be done with the latter as well.
That’s true in the case of hiring outside consultants, said Leach, citing just one example. “We had planned on having some technical consultants come in and assist IT with some upgrades and migrations, and we decided to leverage Copilot instead to guide us through these steps, and we were able to complete a lot of the work on our own without having to engage these consultants.”
Weibel, like Glidden, said there is no shortage of companies pitching new AI products, and the bank is being diligent about reviewing options and deciding which to invest in.
“There are a lot of shiny things in the marketplace at this point,” he told BusinessWest. “We have to see if it’s the right fit for the institution; it’s just like selecting any other system or tool that we would use.”
Leach agreed, but noted that most small and medium-sized banks are beholden to one vendor that is “like the plumber for everything at the bank.”
“And for better or worse, any new AI tool that comes out of left field has to play nice with that vendor and be reasonably priced to sync them up, which often isn’t the case, for us to even get out of the batter’s box and entertain using them.
“We’re less inclined to use AI in forward-facing or client-facing scenarios, and moreso behind the scenes to drive efficiency, and that’s our rallying point,” he added, summarizing the bank’s mindset with the emerging technology. “That, and trying to move beyond experimentation and one-off cases, which everyone has stumbled across, and actually operationalize in a way that really helps the bank.”
Bottom Line
In most respects, banks, like businesses across all sectors, are just getting started when it comes to using AI to change how people work and how they apportion their valuable time.
As noted earlier, the landscape is changing rapidly and constantly, leaving those in the industry to only imagine what things will look like in just a year or two.
As Leach said, AI is not like previous new banking technologies. It is, indeed, a quantum leap forward.
