Great Southern’s loan officers use genetic AI to “chat” with data.

Members of the commercial lending team at Great Southern Bank are using generative AI to “chat” with their data, particularly policy documents and loan documents. You ask questions like “What are the loan-to-value limits for multifamily properties” or “What are the approval requirements for this loan” and get quick answers.

According to Ryan Storey, Director of Loan Operations, using the chatbot, nCino’s GenAI platform, saves hours of time and makes the entire commercial lending department more efficient.

Great Southern, based in Springfield, Missouri, with $5.9 billion in assets and 89 branches across the Midwest, is one of many banks testing use cases for generative AI, a category of artificial intelligence that generates content based on patterns generated that are found in training data.

JPMorgan Chase Employees use OpenAI’s ChatGPT to write emails and reports. Citizens Bank is developing an AI co-pilot to help call center agents answer questions. City introduces GitHub Copilot to all of its developers. Ally Financial uses generative AI to provide post-call summary records in contact centers. These banks are seeking efficiency gains and cost savings through the use of generative AI.

“In many cases, banks are choosing to do what I would call ‘taking out the waste and putting value in it,'” Michael Abbott, Accenture’s global banking lead, said in a statement interview Earlier this year.

For example, he said using genetic AI to summarize customer calls saves about four minutes per call. Mortgage lenders use it to review loan applications against Fannie Mae requirements and identify potential red flags. This type of operational use can provide banks with a 20 to 30 percent increase in efficiency, he said.

According to Alenka Grealish, principal analyst at Celent, the use of generative AI to increase productivity in commercial lending is relatively new.

“It’s a powerful use case and will likely grow in importance,” she said. “Banks are constantly striving to reduce process cycle time, which represents a competitive advantage. Therefore, a generative AI tool will be part of this, which is relatively low-risk – that is, it gets its answers from internal documents – and accelerates the search for employees. “The current wave of early adopters.”

Great Southern has been using nCino commercial lending software since 2018. This software analyzes loan applications and makes loan approval recommendations. Human loan officers review these recommendations and make the actual decision to grant loans.

Great Southern began rolling out nCino’s generative AI bot Banking Advisor in September. To train it, Storey’s team loaded loan documents into Banking Advisor. Additional procedural documents such as training manuals and policy documents were then fed into the system, including the bank’s 185-page credit policy. If someone wants to know what the loan-to-value limits are for home loans, they enter the question into Banking Advisor, which returns information without the user having to know which document contains the answer.

Using such a large language model as a knowledge base creates another place where updated versions of documents need to be stored. Storey said this is not a burden because the bank has a process in place to ensure there is an accurate list of locations where document revisions are stored. Great Southern first had its credit analysts test the system in a sandbox. The entire commercial lending team now uses it, including loan officers and analysts, about 150 people in total.

One challenge is the different ways in which documents and people refer to certain terms.

“I could go and ask what our risk rating is for a particular type of loan,” Storey said. The term “risk class” or “risk rating” may be used in the credit policy. Storey gets around this problem by teaching users to ask a question in different ways and with different terminology, especially if they get incomplete or no answers on the first try. The AI ​​also references its sources, allowing a user to quickly verify the source material if initial answers seem incomplete.

Storey has not experienced Banking Advisor hallucinating – exaggerating or fabricating answers – as is the case with many large language models.

The fee the bank pays to use the tool is easily justified based on the hours saved, Storey said. And he expects to achieve even greater efficiency in the future.

More recently, the bank has added another use case for Banking Advisor – generating reports for commercial loan balances. The generator uses data from nCino and some predefined prompts to generate a narrative paragraph explaining the loan application, the collateral being pledged, or even a background of the borrower and its related entities. This saves the credit analyst time and keystrokes when writing reports that contain information already entered into nCino.

Accenture’s Abbott has seen other banks using generative AI to create credits. He has seen some commercial RMs deploy generative AI co-pilots to help account executives analyze and recommend products such as cash flow management and to help account executives find new customers.

“It’s slowly but surely changing in almost every part of the commercial banking lifecycle,” Abbott said.

Storey’s team just took a first look at another Banking Advisor tool that can review, sort, and file loan documents into the appropriate locations in the electronic document manager in nCino.

“The chatbot knowledge base feature was really just the beginning of what we want to achieve with Banking Advisor,” Storey said. “We believe AI can play a role in improving the way we do business.”