Why We Built Balerion AI

Mortgage lending was once a relationship business. Today, it's an expensive operational gauntlet. The average cost to originate a mortgage now exceeds $12,000, driven not by capital markets losses but by the day-to-day grind from application to closing. 

Loan officers, processors, and underwriters are trapped in endless loops of requests, corrections, and re-reviews to meet compliance demands. Document requests. Income recalculations. Compliance re-reviews. Handoffs that never quite stick. Talented people buried under work that should never reach their desk.

When we started pulling on this thread, we couldn’t let go. We saw how small, missed details in lending applications could quietly disqualify borrowers and cost them the home they’d been working toward. In that moment, it stopped being an idea and became a responsibility: to meaningfully improve how mortgage lending works and help continue to protect access to the American dream of homeownership.

Two Paths, One Problem

We’ve been friends since our undergraduate years at Berkeley, the kind of friendship that survives graduation, diverging career paths, and even attempts to poach one other into rival startups.

Naren spent years at early-stage startups and at Snowflake building data products for fintech and banking customers. Institution after institution showed him the same frustrating pattern: companies sitting on enormous amounts of fragmented data, desperate to unlock its value but unable to connect the dots. Customers defaulted to generic tools because nothing purpose-built existed. Risk slipped through the cracks. Workflows broke down. The costs, to the business and to the customer, were real.

Vishal came at this from the AI frontier. At Ambi Robotics and UC Berkeley's Sky Computing Lab, he focused on a deceptively hard question: how do you build AI that doesn't just assist humans, but can take consequential action on their behalf, reliably and at scale? That question shaped everything he built.

When we reconnected and started comparing notes, a single insight kept surfacing: mortgage origination is one of the most complex, highest-stakes workflows in all of financial services. The cost is immense, the margin for error is slim, the data is sprawling, and the consequences of getting it wrong ripple far beyond a single deal. Yet the AI being applied to it was nowhere close to what the problem actually demanded.

That gap is why we started this company.

What We Kept Hearing

We spent months talking to executives, loan officers, processors, and underwriters at the nation’s leading mortgage lenders. The frustration was ubiquitous.

A missed income discrepancy caught at underwriting, not at intake. An undisclosed active tradeline buried in a credit report. A self-employed income calculation kicked back twice. Each loop adds days. Every day adds cost. Every delay risks losing the borrower.

The existing technology hasn’t kept pace with the industry. Most AI in loan origination is a patchwork of siloed point solutions that don't talk to each other and leave the hardest parts of the workflow untouched. The gaps are where risks slip through, deals fall apart, and borrowers get lost.

This isn't an operational inconvenience. It’s a strategic crisis. And as purchase market competition intensifies, the lenders who can't close faster and cheaper will simply lose.

What We Built

Balerion Loan Intelligence is an end-to-end agentic reasoning engine embedded directly into the origination workflow. The moment a loan enters the loan origination system, Balerion Loan Intelligence immediately identifies key gaps such as missing documents, unsatisfied guidelines, and complex income analyses - surfacing these along with solutions to loan officers and underwriters so they can focus time on high-value decisions and human relationships. Balerion is able to closely follow not only various conforming regulations such as Fannie Mae and Freddie Mac, but also lender-specific Non-QM rules and overlays, unlocking value through customization.

Every signal informs every decision, feedback loops are built in from day one, and the whole is genuinely smarter than the sum of its parts.

The result: problems that would take days to chase down manually get caught upstream instantaneously before they become bottlenecks. No more rework. Faster closes. Materially lower origination costs. Loan officers and underwriters freed to focus on the crucial decisions that drive deals forward. 

We believe the lenders who win the purchase market won't just have better rates or more qualified borrowers. They'll have better operations at a materially lower cost. Our team is building the infrastructure to make that possible.

We're fortunate enough to be backed by world-class investors including Kleiner Perkins, BoxGroup, and Formation, and to be working with leading lenders like FM Home Loans who are using our platform to drive efficiency, uncover new opportunities, and deliver better outcomes across their business. If you're tired of rising origination costs and losing borrowers to back office inefficiencies — let's talk.