U.S. banks focused on small-business lending are using AI agents first to streamline internal work rather than for customer service. The strategy aims to reduce staff workload by identifying potential customer churn in advance and cleaning up data, while boosting the competitiveness of in-person sales.

American Banker recently reported that small and midsize U.S. financial institutions shared examples of how they use AI agents and discussed future tasks at a small-business banking conference.

Michael Desimone (마이클 데시모네), chief banking and lending officer at Citadel Federal Credit Union, cited improved access to data as AI's biggest advantage. He said it removes the need to review huge volumes of reports and memos one by one to understand a company's business conditions and its position in the industry.

AI is also being used to organise data. Gary Class (게리 클래스), global industry strategist at Teradata, said classifying a bank's raw data and extracting required information takes significant time and manpower. He stressed that AI agents are improving the efficiency of that work.

Capitol National Bank is introducing AI into detection work related to anti-money laundering (AML) and the Bank Secrecy Act (BSA). It is also developing a model to predict the likelihood of customer deposit outflows.

The model will be linked to the bank's core computing system to analyse transaction patterns, address changes and whether regular deposits have stopped. It then plans to assess churn risk by customer and send it to the customer relationship management (CRM) system so staff can respond proactively.

The bank's policy is to leave direct communication with customers to people. Cristin Reid (크리스틴 리드), chief executive officer at Capitol National Bank, said, "We have no plans to have AI contact customers directly." She added, "We will improve internal work efficiency so employees can focus more on customer management."

Regulation and internal controls were cited as issues to be addressed. Desimone said he was concerned financial regulation could lag behind technological development, and Reid stressed the need to establish risk management and vendor control systems before adopting AI.

The assessment is that the purpose of using AI at small and midsize U.S. financial institutions is to raise the productivity of existing staff and strengthen customer relationships rather than simply replace workers.