Brett Caines, co-founder and CEO of Lumos Technologies Inc., got his crash course in banking at Wilmington-headquartered Live Oak Bank.
He joined the Live Oak team in 2007 when the early-stage firm was organizing to become a bank. In 2011, Caines became the company’s chief financial officer – a role he held until he left to start Lumos Technologies with Stephen Hayes and Youri Nelson, also former Live Oak employees, in 2021.
At Live Oak, a cloud-based bank that primarily serves small businesses, Caines and his co-founders saw a “market need for … a way to efficiently and effectively assess small business credit risk,” Caines said.
Traditionally, evaluating a business’s credit risk is a “pretty manual process,” he said, with underwriters pulling data from the Small Business Administration and other government agencies.
“We’re trying to expand the way banks assess credit and do it in a very quick, automated way,” Caines said. “So, we’ve developed a couple of predictive models using AI machine learning. We have a historical data set of about 2 million small business loans and how they performed over the past 30 years that we used to train those predictive models.”
Within about 20 seconds, Caines said, the Lumos model can tell a lender the probability of a loan’s default along with the factors determining that result.
“We want to help banks actually make more loans to small businesses but make better decisions on the loans that they’re making,” Caines said.
The company’s products have evolved and grown since its founding, with new offerings that help monitor the status of closed loans, using predictive models to help lenders identify areas of weakness in their portfolios.
The company has also worked to layer new products into its functions, including a process that helps underwriters draft credit memos and pull in industry- and business-specific information.
“We’ve automated all of that through natural language processing, so really what used to take an underwriter somewhere between a half a day and a full day to curate as part of the underwriting process. … We’ve taken probably a day of that work and moved it down to about one minute with the use of AI,” Caines said.
“By saving underwriter time and actual gathering of the data, it gives them more time to then analyze what’s been pulled together,” he added.
Some of the product evolution at Lumos has come naturally, Caines said, but some new products stemmed from customer inquiries about applying existing Lumos products to other parts of the lending process.
“What started as predictive models for kind of one slice of a loan life cycle, we’ve built products that cover a broader spectrum of the loan life cycle,” Caines said, “from origination all the way through the loan being on portfolio until it eventually pays off.”
In the future, Caines said, he would like to develop a business segment that interfaces directly with small businesses, allowing them to get information about their credit risk from Lumos before approaching a lender.vBut the company is taking a cautious approach to growth, only wanting to expand “from a very strong position with a great product set supporting us,” Caines said.
The company has also taken a collaborative approach when working with other financial technology firms. Last fall, Lumos announced an integration agreement with nCino and has partnerships with other technology firms, with more in the works.
“Fintechs, in general, can better serve the financial services industry if, instead of just constantly competing with each other to offer a slightly better than mediocre product,” Caines said, “we combine in a more collaborative way to add value to our respective client bases.”
Meet the 2025 CEA Winners
