VESTA, an AI-native mortgage origination platform, closed a $30 million Series B led by Conversion Capital, with participation from existing customers Pennymac and New American Funding, plus Citi Ventures and Andreessen Horowitz.
The round brings total funding to $85 million and comes as the startup reported 12x year-over-year revenue growth. Vesta has facilitated origination of more than $100 billion in annual loans for lending partners.
Co-founder and CEO Mike Yu attributed the momentum to a fundamental shift in lender readiness. "Demand exploded in the last year," he said. The startup holds under 5 percent market share and plans to "staff up, take the market, and invest in new product lines."
The unit economics are clear: the average U.S. mortgage closing takes 40 days and costs roughly $11,000 per loan, with human labor the primary cost driver and manual review the main timeline bottleneck. Vesta's AI agents automate significant portions of the origination workflow—document review, data verification, condition tracking—allowing lenders to compress timelines and reduce per-loan overhead.
The product architecture reflects a deliberate go-to-market strategy. Lenders initially approve AI agent output in a supervisory capacity. As confidence builds, they graduate to letting agents handle a share of loans independently. Some customers now permit AI agents to make underwriting decisions directly, though lenders retain final accountability and Vesta logs all agent reasoning for compliance and audit trails.
The capability jump arrived via recent AI model improvements. Yu cited Claude Sonnet 4.5 as a watershed: its ability to follow user-configured instructions across multi-turn, complex workflows made autonomous mortgage processing feasible where earlier models could not maintain context or fidelity through the 10-plus-step origination process.
Vesta competes against both entrenched vendors—ICE Mortgage Technology dominates the legacy space—and other AI-native lenders such as Xpanse. The $30 million will fund hiring and a new product line: a personal assistant for lenders to manage task workflows and agent orchestration.
The bet is straightforward: if AI can reliably handle 30-50 percent of origination work within 12-18 months, lenders capture a 20-40 percent per-loan cost reduction and 10-15 day timeline gain. At industry loan volumes, that's a $10 billion-plus TAM opportunity. Vesta's capital burn and customer concentration risk—two large lenders are both investors and users—remain watch points.
