Across every industry, there are countless stories of employees using their phones to access ChatGPT and other generative artificial intelligence (AI) tools far away from the prying eyes of management. But in the case of BankFive Corp., a mutual savings bank and subsidiary of the $1.9 billion BankFive MHC, headquartered in Fall River, Massachusetts, it wasn’t just the employees doing it. “I was using ChatGPT on my phone all the time,” BankFive CEO Anne Tangen admits. “I was anxious to use it in a work setting, [but] I was uncomfortable doing that.”
Tangen and her executive team ultimately decided to use Copilot, the generative AI tool offered by Microsoft. “Copilot was within a safe zone,” she says. “Our IT folks felt comfortable with that risk and the information security part of it.”
It’s a path many financial institutions choose, says Chris Nichols, director of capital markets for SouthState Bank. A subsidiary of $67 billion SouthState Bank Corp., based in Winter Haven, Florida, SouthState’s correspondent division provides services to more than 1,300 banks. Nichols says most have been reluctant to adopt generative AI tools like ChatGPT or Claude, primarily because they don’t want to put their data at risk. But Copilot provides data protection for free, with even more enhanced data compliance and auditing capabilities in its paid enterprise version. And both incorporate large-language models provided by OpenAI, the owner and operator of ChatGPT. “And so, as banks scrambled for a solution, Copilot was easy,” he says. “Most of the banks, as you know, already were in the Microsoft [365] environment.”
According to a 2026 Cornerstone Advisors survey, 49% of banks said they were using some form of generative AI, and that number jumps to 59% for credit unions. Jim Spignardo, director of cloud strategy and AI enablement at ProArch, an IT services and consulting firm, says credit unions are also more likely to opt for Copilot than its competitors. Through its Smart Start program, ProArch works with them to optimize Copilot by identifying use cases, setting up strategies for security and data governance, an AI-use policy, and ultimately an action plan and road map.
“Some start with a small pilot, and then they experiment with a smaller group,” Spignardo says. “Once they’ve kind of become serious about it and start to deploy it to additional members of their organization, that’s when they typically are ready to pull the trigger on getting the enterprise solution in place.”
That is essentially how BankFive rolled out Copilot. Tangen says a small pilot group of about 20 willing employees got the paid enterprise version, which costs an average of $30 per user. But the bank quickly provided access to the free version of the chat tool for all staffers. “I think we wanted everybody to get comfortable with using it as a productivity tool and then let their imaginations take hold and [learn] how else they could use it to perform their job,” she says.
Providing Productivity and Efficiency
Tangen says she uses Copilot to research topics she needs to learn more about, such as stablecoin, or to find out how proposed rules and regulations might impact BankFive. “It just points you in a direction that it would take me months to try to figure out and find the right resources,” she says.
BankFive Executive Vice President and Chief Operating Officer Catherine Dillon says the bank has also used it to help map out initial drafts of policies and procedures, and knows many employees are using it to write more professional emails or create better looking presentations. They then share what they learn with each other. “I think the cross-pollination of what I might be using it for in my job will be different, but it’s like, ‘Oh, I didn’t know [it could do] that. I’m going to check that out too,’” Dillon says. “It’s become its own entity, and people are really having fun with it and are just shocked at how much more efficient they are as a result of using it.”
Spignardo says that is a reasonable approach and much more realistic than what some institutions initially expect. “What typically people want right out of the gate is, ‘We want to connect this through our line of business applications and our Microsoft 365 data, and have all the context be understood and everything to work perfectly,’” he says. “We just have to set the expectations of, ‘Let’s start small before we decide to fly. Let’s learn to walk first.’”
Spignardo also encourages experimentation, but believes it has its limits. ProArch walks its Smart Start customers through the best ways to prompt Copilot to ensure it delivers the most effective responses. “If I am a little bit more clear on what my expectations are, and I provide who the audience [is] for this content I’m generating, I will get a much better output,” Spignardo says.
Dillon understands the idea behind training staff to use it properly, but believes BankFive benefited by allowing employee creativity. “If we had a consultant come in now, people would be like, ‘Yeah, tell me what else I can do!’ They’re excited about it,” she says.
An Eye on the Bottom Line
No matter how financial institutions are deploying Copilot, it is making their employees more productive, says Nichols. He estimates the average bank employee is saving 25 hours per month, which frees them up to do other work. But there’s a catch. “When the CEO wants to know, ‘Well, what has Copilot done for us?’ It’s hard to point [to any cost savings],” he says. “And I think that’s kind of the paradox of what we call point productivity solutions like Copilot.”
To get to a point where Copilot isn’t just saving time but also money, Nichols says it is important for financial institutions to take an enterprise-wide approach. BankFive is still working toward that goal, with some departments slower to adopt Copilot than others. Tangen says her finance and credit departments have yet to use it much, despite her belief it could also help them be more efficient. “In credit, I think you could use it as a credit analytical tool,” she adds. “But I don’t think they’ve used it or see it as a tool that they could take advantage of.”
Nichols says the sooner financial institutions can get all their departments in sync, the sooner they can start using Copilot to advance their bottom lines to make them more competitive. “I’ve had a consistent theme over the last 20 years of saying that banks need to get down to a 40% efficiency ratio to be competitive,” he says. “I believe it has to be lower than that at this point, because of the advent of generative AI. And the only way you’re going to get there is to automate more processes.”
Previous