LendingClub is leaning on AI-powered underwriting to streamline decisioning without much human in the loop. 

The San Francisco-based company’s AI-powered underwriting ecosystem comprises more than 60 production machine-learning models built in-house and more than 250 tailored decision strategies operating across underwriting, pricing, fraud and collections, Senior Vice President and Chief Consumer Credit Officer Kiran Aware told FinAi News.

Most consumer lending decisions can be made in less than five minutes, Aware said, adding that roughly 90% of loans are processed without human intervention, driven by advanced underwriting, fraud detection and document verification. 

It’s not just the speed of AI models but the accuracy that helps LendingClub grow its books, Aware said.  

Highlighting this, LendingClub reported a 40% year-over-year increase in originations in the fourth quarter to $2.6 billion.  

AI a ‘workhorse’ in underwriting

LendingClub’s proprietary model helps it achieve 40% lower delinquencies compared to its peers, Aware said. 

“Two borrowers may look identical by FICO, income and balances, but our system can distinguish a borrower who is actively deleveraging from one rapidly accumulating debt — a difference traditional models often miss,” he said. 

Machine learning, a subset of AI, is the workhorse in underwriting, and LendingClub has used it for well over a decade, Aware said.  

“Beyond that, we’re increasingly using AI to enhance other aspects of our lending business, from onboarding and identity verification to servicing and collections,” he said. “Even when additional documentation is required, recent AI-assisted verification tools have reduced processes that once took days down to minutes.” 

Agentic efforts

Agentic AI is also part of LendingClub’s strategy this year, Aware said. It is becoming a “core part” of LendingClub’s product roadmap.

“We’re already using it in our product development process, and we’re using it to automate some complex workflows.” 

The company plans to deploy agentic AI in multistep complex workflows where AI can monitor signals, take actions and escalate to humans as needed, Aware said. 

Areas where LendingClub aims to deploy agentic AI include: 

• Fraud investigations;  

• Operational decisioning; and  

• Internal analytics. 

The company did not disclose whether it plans to develop its agentic AI tools in-house or with a vendor. 

Register here for the inaugural FinAi Banking Summit, taking place March 2-3 in Denver. View the full event agenda here.