Mortgage lender OCMBC has transferred its AI strategy in-house after several failed projects with vendors. 

The Irvine, Calif.-based lender looked to third parties to develop AI projects, including automated underwriting, but the projects were unsuccessful, Thomas Shaw, chief technology officer and chief marketing officer at the lender, said during the “AI Agents & Lending: Unlock Your Competitive Edge” webinar on Feb. 18 hosted by Multimodal. 

A lot of vendors take a “one-size-fits-all approach,” he said. “That won’t work for us.” 

With the failed AI projects, either the vendor didn’t deliver on promises, timing of costs were off, and “just nothing really aligned,” Shaw said. “So, we stepped back.” 

That’s when the lender turned to its internal development team to build out AI, he said, noting document processing was the starting point. 

“Documents are … the core of everything that comes in,” Shaw said. “We really need to understand what those documents look like, where they go and the data that comes off them.” 

The lender’s development group first worked on document indexing classification, he said, and is moving into the following AI-driven solutions:  

• Document verification;

• Income calculation;

• Review;

• Validation; and  

• Document distribution.  

“It’s a lot to go into, and there are many sub projects getting here, but a lot of the team is very engaged in it,” he said. 

OCMBC ranked 22 out of the 50 largest mortgage lenders in 2025, according to the industry benchmark provider Scotsman Guide. Its origination volume reached $6.5 billion in 2024. Rankings are based on the prior year’s data.  

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