Mortgage origination software startup Vesta has raised a fresh $30 million financing round led by Conversion Capital. The round drew participation from major industry clients, including Pennymac and New American Funding, alongside Citi Ventures and Andreessen Horowitz. This new capital injection lifts the total funding secured by the company to $85 million.
Processing Over $100 Billion in Annual Originations
Founded in 2020 by Devon Yang and Mike Yu, who serves as CEO, the company builds AI-native tools to streamline the mortgage pipeline. Yu noted that client demand surged dramatically over the past year, making this an ideal window to secure growth capital. The business experienced a 12-fold increase in revenue year over year while helping lenders originate more than $100 billion in annual loans. Even with rapid adoption, Yu highlighted that the company currently commands under 5% of the overall market, prompting plans to recruit additional personnel, capture larger market share, and develop fresh product lines.
Tackling the 40-Day, $11,000 Loan Bottleneck
Among the upcoming tools in development is a dedicated personal assistant designed for mortgage originators, intended to execute workflows and track operational milestones. Within the United States, closing a standard mortgage requires roughly 40 days and generates approximately $11,000 in expenses per loan. A predominant portion of that expense stems from human labor, with the most persistent operational delay involving wait times while an application sits in queue for individual manual review.
Deploying Swarms of Agents for Underwriting
To eliminate these friction points, Vesta enables mortgage operations to orchestrate swarms of AI agents to complete underwriting and origination tasks concurrently. Lenders retain full control over which duties are delegated to the automated systems. In typical rollouts, client firms initially set up agents under strict human supervision and sign-off, eventually granting autonomous oversight to a specified share of loans before widening the footprint. Certain lending organizations have even advanced to using these agents for core underwriting decisions. Nevertheless, financial institutions maintain legal and operational responsibility for every underwriting outcome, while the platform maintains detailed audit logs of all actions and reasoning to satisfy compliance standards.
Model Advances Led by Sonnet 4.5
The operational autonomy now possible in lending automation reflects swift advances in foundational artificial intelligence over the past year. Earlier iterations lacked the consistency needed to carry out intricate, multi-step financial tasks. Prior to these developments, Vesta focused primarily on designing resilient data architecture that could support future automation tools. Yu pinpointed Claude Sonnet 4.5 as a decisive breakthrough, noting that it proved significantly superior to previous generations at strictly honoring user-configured guidelines across extended operational horizons.
Taking on Legacy Systems and Modern Rivals
As Vesta broadens its footprint, it faces established industry players like ICE Mortgage Technology as well as newer AI-native software builders such as Xpanse. Addressing traditional infrastructure providers, Yu emphasized that legacy platforms were never engineered around autonomous agent logic, making retrofitting AI on top of them an arduous task. Looking ahead, the company aims to convert the broader mortgage lending ecosystem into active customers before expanding into adjacent financial domains dictated by client demand.
