Vesta, a startup building AI-powered software for mortgage origination, secured $30 million in new funding led by Conversion Capital, the company announced Thursday. The round included investments from three of Vesta's customers — Pennymac and New American Funding among them — plus Citi Ventures and Andreessen Horowitz. Founded in 2020, the company uses AI agents to handle much of the loan origination workflow, aiming to cut the time and expense lenders face when processing home loans.
Vesta has now raised $85 million in total funding. CEO and co-founder Mike Yu told TechCrunch that revenue jumped 12 times compared to the previous year, and demand for the product "exploded in the last year." The company still holds less than 5% of the market. In the U.S., closing a mortgage typically requires around 40 days and costs roughly $11,000 per loan, with most of that expense coming from human labor. A major delay, according to the report, stems from waiting for staff to review individual loans.
"While the traction is great, we are still under 5% market share and now is the time to staff up, take the market, and invest in new product lines," Yu said. The startup's approach lets humans deploy what it calls a swarm of agents to complete tasks more quickly, with customers choosing which jobs to assign the AI. Many lenders begin by having a person approve the agent's work, then allow it to handle a portion of loans independently before expanding its responsibilities. Some clients now use Vesta's AI agents to make underwriting decisions, though Yu noted that lenders remain responsible for those choices regardless of the software involved, and all actions and reasoning are logged for compliance and auditing.
The sharp improvement in AI models over the past year made this level of automation possible, Yu explained. Earlier versions of the technology weren't capable of handling the complicated, multi-step processes involved in mortgage lending. Before that shift, Vesta focused on creating the right data structure to leverage the most advanced tools available. Yu pointed to Claude Sonnet 4.5 as a breakthrough, saying the model proved "much better at adhering to user-configured instructions over the time horizons we need than previous generations." The company now competes with traditional mortgage platforms like ICE Mortgage Technology and other AI-focused startups such as Xpanse. Against legacy systems, Yu argued, Vesta holds an edge because older platforms weren't designed with AI agents in mind, making it "very hard" to layer agents onto them. The new funding will support hiring, market expansion, and development of additional product lines, including a personal assistant for mortgage issuers to track workflows and perform tasks.
Looking ahead, Yu said the company's priority is "earning the business of the rest of the mortgage industry," with plans to expand into whatever areas customers demand. The backing from existing clients signals confidence that AI agents can handle tasks previously reserved for human underwriters, even as regulatory scrutiny of algorithmic decision-making in lending continues to evolve. Vesta's growth trajectory suggests that mortgage lenders are willing to hand over significant operational control to automated systems when cost and speed pressures mount. The question for financial institutions will be whether efficiency gains outweigh the reputational and oversight risks that come with surrendering judgment to software. As more lenders adopt agent-based workflows, the industry may face a choice between competing on technology or falling behind competitors who've already made the leap.