MikeYu

The Problem He Couldn't Unsee

Walk into most mortgage companies in America and you'll find software that was designed before the web browser existed. Loan origination systems from the 1990s, patched and re-patched, running the financial lives of millions of homebuyers. Origination costs have doubled from roughly $4,000 in 2009 to $8,000 in 2019. A mortgage still takes 45 days to close.

Mike Yu noticed this problem the same way you notice a cracked windshield - once you've seen it, you can't stop seeing it. He joined Blend in 2016 as one of its earliest product managers, still a student at Stanford completing his BA in Economics and MS in Computer Science. He stayed four years. In that time, he helped land the Wells Fargo partnership, co-built Blend's Day 1 Certainty product with Fannie Mae, and founded Blend's insurance agency division from scratch - including becoming the company's first-ever licensed insurance agent just to understand the regulatory stack from the ground up.

That hands-on, go-get-licensed-yourself approach isn't a quirk. It's the operating principle. Before writing a single line of Vesta's code, Yu and co-founder Devon Yang - his colleague from the Blend years - spent the better part of 2019 and 2020 interviewing lenders. Not a focus group. Not a survey. More than half of the top 500 U.S. mortgage lenders, one conversation at a time. The founding hypothesis that emerged from those conversations has not changed since day one.

A Different Kind of Founder Path

Yu's pre-Blend resume is the kind you only get when you're genuinely curious about how things work. He wrote software at Bloomberg, traded at Jane Street Capital - one of the most selective quantitative firms in the world - and worked on product and special projects at Nearpod, an edtech startup. The common thread isn't finance or tech. It's systems: how they work, where they break, what it would look like to rebuild them from scratch.

At Stanford, he also co-architected Finprint - a blockchain-based secure financial profile system designed to put consumers in control of their own financial data. It was open-sourced after development. This wasn't just a side project; it seeded a relationship with Conversion Capital that eventually led to Yu joining the firm as an investor before spinning out to build Vesta full-time.

The Bet on Infrastructure

When Vesta launched in November 2020, it did so at the worst possible moment for anyone trying to make a calm, long-term bet on mortgage infrastructure. Pandemic-era refinancing demand was hitting all-time records. Every lender in America was overwhelmed. And that, paradoxically, was precisely the right time.

Vesta's architecture is built around three core ideas Yu developed from those 250+ lender conversations: a modern data management system designed for mortgage information from scratch, a no-code workflow automation engine with a rules-driven task model, and an open API platform that treats the LOS as infrastructure - a neutral layer that lets lenders pick best-in-class point solutions rather than being locked into a monolithic vendor's ecosystem.

$55 Million and a16z's Conviction

Conversion Capital and Bain Capital Ventures led Vesta's $5 million seed in late 2020. In January 2022, Andreessen Horowitz led a $30 million Series A - with a16z General Partner Angela Strange calling the LOS gap "the consistent engine to orchestrate and standardize loan origination" that the entire industry was missing. Zigg Capital joined that round alongside the returning seed investors. Total funding reached $55 million.

Strange's framing was precise: "It requires the technical and financial know-how to develop and implement a new backbone in a highly-regulated industry. All the players agree that a new system is needed, but to date no one has successfully built it." That's the market Vesta is attempting to own.

Playing the Long Game

Yu is candid about the difficulty. Replacing a system of record - the core database that every loan decision flows through - is not a natural "wedge" into a market. You can't slip in quietly via a point solution and gradually expand. You have to convince a lender to swap the beating heart of their operations. That's change management as much as product management.

His pitch is architectural: the old systems weren't built wrong for their time. They were built for a world before cloud computing, before modern APIs, before AI could read a tax return and flag a discrepancy in seconds. Vesta isn't just a faster LOS. It's a reimagining of the atomic primitives - the fundamental data structures and workflow units - that software can now orchestrate in ways humans cannot.

Recognition and the Road Ahead

The industry has been paying attention. Yu was named a 2020 HousingWire Rising Star, then a 2024 HousingWire Vanguard. Forbes included him in its 30 Under 30. The Generalist named him to its Future 50. Vesta took a spot in the 2025 HousingWire Tech100. In December 2024, Bloomberg featured him in a segment on why a one-click mortgage remains so stubbornly out of reach - and what it would actually take to build the infrastructure that makes it possible.

His answer, predictably, starts with the LOS. If the core system of record can be opened, modularized, and run on AI-native infrastructure - the kind that routes tasks, checks compliance, and flags anomalies in real time rather than at the end of a 45-day review cycle - then the downstream experience for borrowers changes fundamentally. The one-click mortgage isn't a UX problem. It's an infrastructure problem. And Mike Yu is three years into what may be a decade-long bet that he can solve it.