Hatch Bank is taking an inclusive approach to identify effective AI applications and propel its innovation strategy.  

The digital bank conducts weekly check-ins during which employees describe how they’ve used AI, President Amanda Swoverland told FinAi News.  

This approach has enabled the $182 million bank to maximize practical AI tools such as Microsoft Copilot and LLMs, she said.  

For example, employees use Copilot to search for specific emails that may be buried in their inbox, she said, adding that they can also use Claude or ChatGPT to quickly learn spreadsheet formulas. 

“Little tasks that used to take you five or 10 minutes to do, you can now do in seconds,” Swoverland said. “We’re really leaning into how we can use AI in every corner of the organization.” 

Frequent check-ins are spurring broader AI initiatives by improving prompts engineering, with Hatch Bank using agentic AI tools for document review and compliance processes, she said. 

Loan sampling is one example of this, Swoverland said. 

“If you write the right prompt and you have the right data, you can start to use AI agents to go in and start curating and flagging things for you.”

— Amanda Swoverland, president, Hatch Bank

“You can do a full-scope review and start looking for anomalies in ways that allow you to go deeper and broader within your database,” she said.

Loan sampling is a risk-management practice in which banks evaluate a subset of loans to determine whether they comply with fair lending and other regulations, according to advisory and consulting firm Cherry Bekaert.  

Hatch Bank also bolstered compliance operations in January when it deployed fintech Themis’ AI tool for complaint management.  

Digital banks’ AI needs 

Digital banks have specific AI needs due to the speed and quantity of transactions, Swoverland said. 

For instance, the average approval time for a small business loan is one to three days on average for digital lenders, compared with three to eight weeks at traditional banks, according to commercial lender Crestmont Capital.  

“You need the tools that help you to really wrap your arms around the data that you have and how you can scale,” Swoverland said. “If I have a tech company that can help me scrub 1,000 [transactions] and do it in a minute and do it, quite frankly, better than a human set of eyes because it can cross analyze all the different data patterns, that’s what I want.” 

Digital banks also value “integration layers” that connect AI models and enterprise data, Swoverland said. 

“I would rather buy one software and pay a little bit more for it, but have it do all of the things as one central hub,” she said. 

Register here for the FinAi Lending Summit, set for Oct. 7-8 in Las Vegas.