Unlocking data with Blend Vision

Talk to any industry expert today, and you’ll likely hear the term “margin compression.” It’s becoming more difficult for lenders to keep origination costs down, and many are turning to technology for solutions.

At Blend, we believe in technology’s ability to fix the problems our customers face. In an effort to win efficiency gains, we’ve been investing in building functionality that allows loan teams to focus on high-value activities instead of time-consuming manual ones like stare-and-compare.

Looking from paper documents to a computer screen to paper documents is the reality today for many loan teams. Most loan data is trapped in manually uploaded paper documents and thus, software vendors have traditionally been unable to read in an entire loan file and intelligently trigger workflows based off it.

So, how do we solve for this?

Introducing Blend Vision

Our new functionality, Blend Vision, is the first step along a much longer journey to get the most out of data trapped in uploaded documents.

While Blend already leverages digitized, source-verified data wherever possible, the reality of the lending process is that it is heavily dependent on paper documents. Even with 70 percent of borrowers opting to connect directly to source data in the Blend application, 100 percent of Blend loans include at least one uploaded document.

Instead of relying on error-prone stare-and-compare techniques to condition a loan, loan teams will be able to unlock data from uploaded documents through computer visioning and take advantage of Blend’s automated follow-up workflow, which allows borrowers to satisfy conditions with 50 percent fewer interactions. This more intelligent and efficient experience is one way Blend helps lenders keep origination costs down.

Why Blend Vision is more than OCR

Blend Vision is different from other optical character recognition (OCR) solutions because it is native to the point-of-sale system and comes with quality controls.

Finding, signing with, and managing a technology vendor — outside of an existing technology stack — is costly and time-consuming. It requires human resources to manage a separate vendor and can introduce friction between different platforms. Lenders are already struggling to juggle technology vendors to keep up with the pace of innovation. Blend Vision provides an end-to-end solution where computer vision is baked into the core point-of-sale experience and customers are able to leverage the benefits from day one.

Lifting data from documents is critical functionality but it should never come at the cost of data quality and accuracy. At Blend, we take data integrity very seriously; we believe that data quality is much more important than data quantity. To ensure data accuracy, Blend Vision includes controls to guarantee quality. These checks include automated data validation, real-time monitoring, and manual quality control.

Automated data validation constantly ensures that the data lifted from uploaded documents is well-formed and complete. If data does not pass our validation check, then these fields will not be used. Real-time monitoring allows us to track trends in our computer visioning success rate. Whenever we see anomalies in the trendlines, we manually spot-check documents to ensure good coverage. Lastly, we routinely perform manual quality checks by comparing documents and data inputs as another layer of diligence. Like everything else we build, Blend Vision is built to bring integrity to the data our loan teams use to fulfill conditions and fund loans.

Combatting margin compression with efficiency gains

Today, Blend Vision is able to parse certain bank statements and identify large deposits. In the future, we hope to unlock data from all types of documents used in the mortgage process. Our goal is to digitize every document and use the data that’s been trapped to identify when the wrong document has been uploaded, flag data discrepancies, and power more efficient follow-ups.

This new technology will translate to huge efficiencies, additional days shaved off of loan funding time and decreased loan origination costs. While paper documents will likely comprise a large percentage of mortgage applications over the next several years, Blend Vision will empower lenders to review the documents efficiently and accurately.

Read more about our product here, or check out opportunities to join our teams in San Francisco and New York.