Turning community bank employees into AI enthusiasts

Community banks find their employees can be hesitant when the institution integrates artificial intelligence into its operations and workflow, whether it's fear that jobs will be taken away or uncertainty about what to expect. 

Getting them on board can be done in a variety of ways, according to a series of panels on Thursday during an Independent Community Bankers of America forum. Sometimes enthusiasm becomes infectious when employees experience promising use cases firsthand, receive proper training or observe successful uptake among customers. Other times they need reassurance that their jobs will remain, and become more efficient

Community banks are in the early days of experimenting with AI and the earlier days of writing policies to govern its usage. It's key to do both at once.

May 19
Left: Merchants & Marine Bank branch. Top right: Ryan Hildebrand, chief innovation officer of Bankwell Bank. Bottom right: Kim Kirk, chief operations officer of Queensborough National Bank & Trust.

Trial periods or gradual adoption can also give bankers time to ease in. 

When Mercantile Bank in Grand Rapids, Michigan, turned on live chat from AgentIQ for its small-business customers in December 2023, which was after the bank introduced the feature for retail customers, bankers "were freaking out," said Autumn Redman, retail operations manager at the $5.5 billion-asset institution, at the forum. "It's the what-if, 'we don't know what the customers are going to ask.'"

Chat hours initially lasted from 9 a.m. to 5 p.m., but Redman later extended them to match the contact center's hours of 8 a.m. to 6 p.m. "Easing in quietly for your staff helps … and makes sure they have the tools to handle what you need," she said. The bank also took three months to go live with AgentIQ, a month longer than the two months AgentIQ says is typical after signing a contract. 

"We wanted time to fully train our bankers on what we felt they needed," she said.

Heritage Bank in Erlanger, Kentucky, turned to HuLoop Automation, an AI-powered process automation platform, at the end of 2023. Chief Operating Officer Kevin Mooney was drawn to HuLoop's focus on community banks and its affordability compared to other products. The $1.7 billion-asset bank has since built out five automations. 

"The thing I like about the platform is we don't have one or two big problems," he said. "We have 40 or 50 little problems," which IT employees trained on HuLoop's system can build themselves. 

Mooney acknowledges some pushback from operations staff. 

"The first thing people think once we start automating is oh my gosh, you're going to get rid [of us]," said Mooney. 

While the bank has not backfilled vacant positions, "we're not pushing anyone out," he said. "It's been nice for our people to work on things that focus on customers."

At Coulee Bank in La Crosse, Wisconsin, Director of Technology and Information Security Brandon Pelinka turned to an AI assistant from compliance firm Finosec that helps banks prepare for exams by analyzing data from regulatory handbooks.

It addressed a palpable problem. When he joined Coulee, "The IT landscape was somewhat chaotic," he said. The bank had invested in technology, "but much of it was not being used effectively. From a governance perspective we had significant challenges." After experiencing his first exam at the $568 million-asset Coulee, implementing Finosec's exam readiness tool became a priority. 

"I've never been so excited about a product," he said.

Trials also give employees time to adapt to a new way of working.

Bankwell Bank in New Canaan, Connecticut, piloted a new use case with Senso, a firm that uses AI to turn unstructured policies and other raw data into what it says is a cohesive format. While Senso's platform was "already successful" on the retail side, said Senso Co-founder and CEO Saroop Bharwani, standardizing policies and procedures for small-business lending was new. 

Ryan Hildebrand, chief innovation officer at Bankwell Bank
“When this new technology comes online, some folks ... roll up their sleeves and jump into it,” said Ryan Hildebrand, chief innovation officer at Bankwell Bank. “Some of the other employees at the bank may be nervous."

"LLMs [large language models] let us identify gaps and conflicts in policies and make improvements," said Bharwani. The pilot involved ingesting standard operating procedures, or SOPs, issued by the Small Business Administration and having Ryan Hildebrand, the $3.1 billion-asset institution's chief innovation officer, and Dhruva Patel, the bank's small-business lending credit manager, spend three weeks determining how accurately it captured information from the SOPs.

Bankwell's SBA program is "not massive," said Hildebrand, but it's an area the bank wants to ramp up. Patel found that when asking generic questions, such as what rates are allowed by the SBA on certain dollar amounts, the responses were accurate. More technical questions about specific scenarios, such as whether a business owner selling 85% of their business can stay on as owner, returned accurate responses 80% to 90% of the time.

"When this new technology comes online, some folks like Dru roll up their sleeves and jump into it," said Hildebrand. "Some of the other employees at the bank may be nervous. But they are going back and double-checking and realize it's the right answer. It's starting to build into the culture that we can trust this stuff."

For reprint and licensing requests for this article, click here.
Artificial intelligence Community banking Technology
MORE FROM AMERICAN BANKER