Vesta, a startup developing AI-powered software to help lenders originate mortgage loans, announced that it had raised $30 million in a round led by Conversion Capital. Three of the company’s customers participated in the round, including Pennymac and New American Funding, along with Citi Ventures and Andreessen Horowitz.
Vesta was founded in 2020 by Mike Yu and Devon Yang. Yu, who serves as chief executive officer, said the company’s revenue had grown 12-fold year over year and that it had raised $85 million to date. According to the company, its market share remains below 5%, making team expansion and investment in new product lines the focus of the next phase.
Agents to Execute Loan Origination Stages
The platform uses AI agents to automate large portions of the loan-origination process, which takes about 40 days in the United States and costs approximately $11,000 per loan, according to figures cited by Yu. Vesta’s founders believe that a significant portion of the cost is tied to human labor, while one of the main bottlenecks arises from waiting for a file to be reviewed by a specialist.
The company is working to add a personal assistant for mortgage loan originators that can execute tasks and track workflows. Customers decide which tasks the agents can take over; some begin by using an agent under the supervision of an employee who approves its outputs, then gradually expand its scope so that it processes a larger share of loans independently. Yu said some lenders also use Vesta agents to make underwriting decisions.
Human Oversight and Auditing Are Part of the Design
Yu emphasized that responsibility for underwriting decisions remains with the companies, regardless of the software or AI agents they use. According to him, the platform records all actions and the reasons behind decisions, with the aim of supporting compliance requirements and enabling audits of decisions made by AI.
Yu attributes the current level of autonomy to improvements in AI models over the past year. He said Vesta found Claude Sonnet 4.5 better than previous generations at following user-defined instructions over extended periods, an important capability for multistage tasks in mortgage lending.
What Is Changing in Practice?
Vesta does not offer limitless automation, but rather a gradual model for transferring tasks from employees to agents while keeping decision-making and responsibility with the lender. The importance of the round is not tied solely to the funding, but also to the growing reliance of a sector highly sensitive to compliance on agents capable of executing long workflows and documenting the rationale behind their actions.
At the same time, underwriting decisions and human oversight remain critical points, and the company’s ability to scale will depend on how willing lenders are to accept this level of autonomy. Vesta competes with traditional mortgage systems such as ICE Mortgage Technology and other startups, including Xpanse; Yu believes that legacy systems were not originally built to run AI agents on top of them.