Upstart uses AI and more data than in typical credit decisions to dig deeper into a borrower’s true risk and to match customers with lenders who believe in its underwriting assessments.
Rouse said the company’s AI has led to approvals of customers incorrectly perceived by the market as too risky and avoidance of loans customers incorrectly perceived as safe. This increased accuracy has led to approval rates rising 43 percent and interest rates falling 43 percent in the company’s personal loan business, according to Upstart.
Upstart automates all its credit decisions, according to Upstart CEO Dave Girouard, but it still might need to verify customer information. While AI helps here as well, only about one-third of indirect auto loans are truly automated without any additional customer stip activity, according to Rouse. But he said this proportion could grow, noting that Upstart’s personal loans were fully automated 75 percent of the time.
“That’s where we think we’re headed,” Rouse said.
Girouard said he thought the rate could even be higher for auto loans, as the risk of fraud in that sector is reduced by dealers’ ability to see the loan applicants in person.
Rouse said Upstart’s digital retail and lending services allowed for the addition of finance-and-insurance products to a deal.
“One of the value propositions of the AI-powered financing is we are able to approve people for larger loan amounts,” he said.
Girouard saw a role for Upstart in helping automaker-affiliated captive finance companies capture more volume with reduced losses, and he said the company has had discussions with those automaker-affiliated lenders.
“I think that’s something we would hope for in the future,” he said.
