Vesta Raises $30M Series B for Mortgage AI Agents

Before an AI agent can process a lender's next mortgage, the lender has to teach its software how the business works. Vesta has spent nearly 6 years building a loan origination system for that relationship, and its newest financing puts the work of adopting that system close to the center of the story.

The San Francisco mortgage technology company announced a $30M Series B on October 8, 2026, led by Conversion Capital. The financing brings Vesta's total funding to $85M and supports its effort to scale AI agents across mortgage origination. Co-founder and CEO Mike Yu says the company also needs to help lenders absorb its technology faster.

The commercial significance reaches beyond adding an agent to a loan file. Mortgage lenders buy an operating environment that has to carry their procedures, route their exceptions and preserve a record of decisions. Vesta's opportunity now includes making that environment easier to put to work across another institution.

Vesta's $30M Series B brings customers into the financing

Vesta names Pennymac, New American Funding, Citi Ventures, nbkc bank, First American and FirstKey Mortgage as strategic participants. Andreessen Horowitz, Zigg and Navitas also participated. The announcement places institutions with direct reasons to care about mortgage operations alongside financial investors, connecting the vendor's development plans with buyers' daily work.

That relationship has a documented history at Pennymac. The lender announced its Vesta partnership in September 2025, including a long-term minority equity investment. Pennymac said Vesta was live in its Consumer Direct channel, with plans to extend the technology across additional channels. The customer relationship therefore comes with both operating exposure and a financial interest in the software company.

The earlier capital needs careful accounting. Conversion Capital's 2022 investment account separates a $5M Seed round from a subsequent $30M Series A. Vesta's public launch described $35M raised, while a March 2025 Fortune interview reproduced on Vesta's site reported another $20M from strategic investors. Those amounts reconcile with the $55M preceding this Series B and the newly announced $85M cumulative total. Vesta has not disclosed a valuation for the latest round.

The mortgage system has to accommodate people and agents

Yu and co-founder and CTO Devon Yang started Vesta in 2020 after working at Blend. Vesta's founding account describes lenders asking for a better system behind the mortgage application experience. The founders chose to rebuild the core loan origination software, a project that required approximately 2 years of platform development before lender conversions, according to the latest announcement.

Vesta's loan origination system lets people work on the same loan concurrently, routes tasks and gives administrators controls over rules and checks. Its AI agents operate within that environment, with permissions, human escalation and recorded actions. The useful architectural question is whether a lender can assign work, recover the reasoning and intervene through the same system its employees already use.

Yang's explanation of the architecture separates predictable rules from model-based interpretation and workflow coordination. That distinction matters in a mortgage file, where calculations coexist with inconsistent documents and exceptions. The company describes software that combines these methods; its product materials do not establish a blanket guarantee of accurate underwriting or compliance for every deployment.

Customer measurements give the operating argument substance

Pennymac's SEC-filed Q4 2025 earnings presentation reports approximately 50% efficiency gains for loan officers and approximately 25% less loan processing time in its Vesta comparison. The presentation shows average processing time falling from 14.5 hours to 11 hours. These are customer-reported results from a specific implementation, with Pennymac's minority investment disclosed in the same presentation.

Vesta's October announcement reports more than 12x revenue growth over the preceding 12 months. It also says lenders already using the platform or migrating to it collectively originate more than $100B annually. That lender volume includes migrations still underway, and Vesta does not disclose an absolute revenue figure alongside the growth multiple. The measures describe different parts of the business and should remain distinct.

The company says about 40% of completed tasks on Vesta now run through AI agents and automated workflows together. Its agent product page describes staged adoption: start with a bounded task, supervise the work and expand its responsibilities. For an operating team, that creates a way to change the assignment of work while retaining a place for experienced people to review exceptions.

DevCuration’s coverage of Valon’s mortgage servicing software, Elio’s AI-native mortgage brokerage and Sela’s mortgage voice agents follows different parts of the lending workflow. Vesta’s position is the core origination system where those operating decisions and loan records come together. That makes its implementation work a distinct part of the mortgage AI investment story.

Faster implementation becomes part of the product ambition

Yu identifies a practical constraint inside Vesta's own expansion: translating a lender's procedures into thousands of settings can take most of a year. The company wants agents to learn those procedures through documents, questions and historical loans. That proposed onboarding method is a development direction, distinct from the agents Vesta says are already doing production work.

The distinction also explains why this financing has an implementation story. A larger customer pipeline still needs people and tools capable of converting each institution's operating knowledge into a reliable configuration. Work that shortens that conversion could let Vesta's transformation team spend more time helping lenders reorganize responsibilities around the software. That is an operating implication of the plan, rather than a measured result of a released onboarding agent.

Mortgage teams adopting Vesta will continue to decide which tasks can run automatically, which exceptions deserve attention and what evidence a reviewer needs. As the company expands, those decisions travel from one lender's procedures into a working loan file, then into the next review. The Series B gives Yu and Yang more resources to make that passage easier for the institutions waiting to move.