\n<\/p>\n
Vesta<\/a>, an AI-native software startup that helps lenders originate mortgages, announced on Thursday that it raised a $30 million round led by Conversion Capital.<\/p>\n The startup uses AI agents to automate much of the loan origination process, which, the founders say, helps lenders reduce the time and costs of processing mortgages. Three of the company\u2019s customers, including Pennymac and New American Funding, also invested in the round alongside Citi Ventures and Andreessen Horowitz.\u00a0<\/p>\n Mike Yu co-founded the company in 2020 with Devon Yang. Yu, the company\u2019s CEO, told TechCrunch that now was the perfect time to raise because demand for the product has \u201cexploded in the last year.\u201d He said revenue is up 12x year over year and that the company has helped originate more than $100 billion a year in loans for lenders.\u00a0The company has raised $85 million in funding to date.<\/a><\/p>\n \u201cWhile 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,\u201d Yu said.\u00a0<\/p>\n Those new product lines include a personal assistant for mortgage issuers that can help perform tasks and track workflows. It takes around 40 days to close a mortgage in the U.S., costing around $11,000 per loan. \u201cMost of that cost is human labor, and a major bottleneck in the timeline is just waiting for a person to get to reviewing your loan.\u201d\u00a0<\/p>\n With Vesta, the idea is to allow humans to deploy a swarm of agents to speed completion of tasks. Customers decide what tasks they want to assign the Vesta agents: \u201cMany of our customers start an AI agent with a person approving its work, then let it handle a share of loans on its own, then expand,\u201d Yu explained, adding that some lenders are even using Vesta AI agents to make mortgage underwriting decisions.\u00a0<\/p>\n Companies remain responsible for underwriting decisions, he said, regardless of what software or AI agents they use, and Yu added that all actions and reasoning behind a decision are recorded for compliance and to audit AI decisions. <\/p>\n This level of autonomy stems from the vast improvements in AI models in the past year, which weren\u2019t good enough, at that time, to build agents for the complex multi-stage tasks involved in mortgage lending. Before, the company was just focused on building the right data architecture to use the most advanced tools to automate the mortgage process.\u00a0<\/p>\n \u201cFor us, the big breakthrough was [Claude] Sonnet 4.5, which we just found to be much better at adhering to user-configured instructions over the time horizons we need than previous generations,\u201d Yu said.\u00a0<\/p>\n Vesta is, in many ways, competing against both traditional mortgage systems like ICE Mortgage Technology and other AI-native companies also trying to automate the mortgage lending process like Xpanse<\/a>.\u00a0<\/p>\n At least against legacy incumbents, Yu said Vesta\u2019s advantage is that those companies weren\u2019t built for AI agents, and \u201cputting AI agents on top of them is very hard,\u201d he said. As for what\u2019s next, \u201cour priority is earning the business of the rest of the mortgage industry,\u201d Yu said. \u201cThen, we\u2019ll go wherever our customers take us.\u201d\u00a0\u00a0<\/p>\n<\/div>\n When you purchase through links in our articles, we may earn a small commission<\/a>. This doesn\u2019t affect our editorial independence.<\/em><\/p>\n