PomegraVesta raised a $30 million Series B led by Conversion Capital, with three of its lender customers investing, to expand its AI-native loan origination system.
- Vesta raised $30M in Series B funding led by Conversion Capital, bringing total capital raised to $85M.
- Customers New American Funding, PennyMac and NBKC Bank invested alongside Citi Ventures and FirstKey.
- Vesta reports revenue up 12x year over year, with lenders originating more than $100B annually on the platform.
Lead
Vesta, the U.S. developer of an AI-native loan origination system (LOS) for mortgage lenders, announced a $30 million Series B on October 8, 2026. Conversion Capital, which backed the company at seed stage, led the round. The new money lifts total funding to $85 million. Valuation was not disclosed.
The company says lenders on its platform originate more than $100 billion in loans a year. CEO Mike Yu said demand has "exploded in the last year," with revenue up 12x from a year earlier. Vesta did not publish absolute revenue, so the multiple has no stated base.
Who Invested in the Round?
Conversion Capital led, and the roster reads like a customer list. New American Funding, PennyMac Financial Services and NBKC Bank, all existing Vesta clients, put money into the round. Citi Ventures, FirstKey Mortgage and Navitas Capital also participated.
Returning backers include Andreessen Horowitz, which led the $30 million Series A in January 2022, along with Zigg Capital and Parker89, the venture arm of First American Financial. Customers buying equity is a useful signal on product fit, though it also ties investors' commercial interests to the company's fortunes.
What Does Vesta Actually Sell?
Vesta sells a loan origination system in which people and AI agents work the same loan file inside one platform. The agents handle document review, underwriting tasks, condition clearing and closing package reviews. Roughly 40% of tasks on the platform now run through AI agents and automated workflows, according to the company's announcement.
Lenders can begin with human approval on every agent action and widen autonomy as they gain confidence. The pitch is aimed at a costly process: closing a U.S. mortgage takes about 40 days, and the average loan costs roughly $11,000 to produce, most of it labor.
Yu has pointed to Anthropic's Claude Sonnet 4.5 as the model that made the approach workable, citing better adherence to user-configured instructions. That dependence on a third-party model is a supplier risk for any application-layer company, though it also means Vesta's gains scale with model quality it does not have to build.
How Large Are the Reported Results?
PennyMac reports a 25% reduction in operational origination costs and about 50% efficiency gains for loan officers after adopting Vesta. Both figures come from a customer that is now also an investor, so they merit independent confirmation as more lenders report. They are still specific enough to test against rival systems.
Vesta was founded in 2020 by Yu and Devon Yang, both formerly of Blend. Its customer base includes PennyMac, Upstart and NBKC Bank. Yu has put Vesta's share of the market at under 5%, which frames the round as a bet on displacing incumbents rather than defending a position.
What Does This Round Say About the Last One?
The Series B arrives nearly four years after the Series A, a long gap for a venture-backed software company. The 2022 round funded a conventional cloud LOS in a market then contracting under higher interest rates and low origination volume. The company's growth story now rests on AI agents layered onto that base, and the 12x revenue figure suggests the second phase has been much faster than the first.
At $30 million, the round is modest by current AI funding standards. It matches the size of the Series A and suggests capital is not the main constraint. Distribution into lenders' production workflows is.
Competitive Context
Vesta competes with legacy systems from ICE Mortgage Technology, which dominates the LOS market, and with AI-native entrants such as Xpanse. Mortgage lenders replace core systems slowly, because a failed migration can halt loan production. That caution favors vendors that can run alongside existing processes before taking over.
The company plans to spend the proceeds on product development, market expansion and hiring across engineering, implementation and go-to-market roles. Implementation headcount in particular points to the work of onboarding lenders onto a new core system.
What Comes Next for Vesta?
Vesta's next test is converting investor-customers' results into wins at lenders with no financial stake in the company. A higher share of tasks handled by agents, now about 40%, would show whether the human-approval model is graduating to greater autonomy. Mortgage volume also matters: a rebound in origination would stress-test the platform's capacity, while a soft market would sharpen lenders' appetite for cost cuts.
Outlook
Vesta enters its next phase with $85 million in total funding, a lender-heavy cap table and a claimed 12x revenue jump from a small base in a market where it holds under 5%. The open questions are valuation, absolute revenue and whether customer-reported savings hold across lenders outside its investor group. Expect incumbents to respond with their own agent features as AI-native systems gain share.