Today, we’re announcing our $30 million Series B, led by Conversion Capital, with strategic participation from Pennymac, New American Funding, Citi Ventures, nbkc bank, First American, and FirstKey Mortgage, as well as participation from a16z, Zigg, and Navitas. This brings our total funding to $85 million.
In the last twelve months, revenue has grown more than 12x, and both customer count and headcount are up more than 50%. Lenders on or moving to Vesta already originate more than $100 billion in loans per year.
The fresh round of financing brings with it the obligation to execute even more aggressively on our mission to automate mortgage origination from end to end.
Lots of people argue that heavily regulated industries, like mortgage lending, will adopt AI last. It’s reasonable for an outsider to assume that; they look at this industry and see thousands of pages of regulation, 20-person risk committees, sales cycles that are traditionally measured in years, not months, and worst of all, 30-year-old systems of record sitting in the middle.
At Vesta, we’ve moved the mortgage industry to the front of the line for agentic transformation. Once you understand the space, it’s clear that it’s perfect for agents:
This isn't just a forecast. Our agents have been in production all year — scrubbing applications, making underwriting decisions, and reviewing closing documents. About 40% of tasks completed on Vesta are now handled by AI agents and automated workflows, and that share is climbing.
Getting in position to make this happen wasn’t easy. AI agents have a huge advantage when they are native to the lender’s core software platform, and building and implementing that platform has been a journey of almost six years already.
When Devon and I started Vesta, the system most of this industry ran on (and still runs on!) was already over twenty years old. It has a different backend field for "loan amount" on every screen the value appears on, so lenders write automated triggers to keep the loan amount in sync with the loan amount. Its database is single-write: one party — human, machine, or integration — can edit a file at a time. Building agents on top of this would need tools to wait for the human to leave the loan, computer use to go in through the UI of a Windows application, and mirroring a bloated data model into a huge, context-inefficient ontology.
We knew the only way to position this industry for what was coming in AI was to rebuild the entire system from scratch. Agents only make that case stronger. Auditability, guardrails, handing work back and forth with a human, navigating a ruleset that runs to thousands of pages — all of it depends on the system of record giving the agent the right tools, and that only happens when the two are built together.
So we spent two years writing the core platform, then converted some of the largest lenders in the country and made sure they loved the results. Pennymac has already reported 25% back-office efficiency gains, with a roadmap to automate 80% of its production workflows by the end of 2027.
This is a transformation business, and the hard work in front of us is scaling up our technology and impact, and helping lenders absorb what we've built faster and faster. The bottleneck right now is us, and there is a lot of software and company to build to make that faster.
Standing up a system of record today means translating how a lender works into thousands of settings, one at a time, by hand, and it takes the better part of a year. That’s not how lenders bring on a person, and it isn't how this industry is going to bring on agents a year from now. Instead, the agent will learn by reading the procedures, asking questions, and running through old loans. What it learns then becomes the rules and workflow that future agents run on. Once configuration is no longer the hardest part of changing systems, our transformation team can spend its time teaching each lender what their operations can look like when their people unblock and manage agents instead of keying data from documents
Even at our current scale, less than 5% of the market is on Vesta. The mortgage market is not immune to the AI revolution: every single lender is going to need agents manufacturing their mortgages. We're going to be the ones who get them there.
When every mortgage origination is automated, borrowers will wait 7 days to close (the minimum required by law), not 45. Loans that aren’t always economically viable to make today — small-balance loans, manufactured homes, incremental refis — suddenly work. Starting a lender will entail setting up some software, not hiring 30 people and getting to critical mass, which means more competition, more choice, and more accessibility.
Someday, every mortgage is going to be manufactured by agents. Our customers are going to originate the first trillion dollars of those mortgages on Vesta in the next 5 years.
A huge thank you to everyone who made this possible:
If you run a lender and want agents manufacturing your loans, talk to us. If you want to help us retool one of the biggest industries in America, we're hiring.