As any successful lender will tell you, accurate data is key to underwriting—and the more of it you have, the better. But incorporating that data into an underwriting system can be operationally challenging at scale. While bank-sync (sharing personal information by providing bank login info) is gaining ground, the vast majority of underwriting data is still provided through paper documentation (business P&Ls, bank statements, paystubs, tax records, etc.).

As a result, today’s lenders employ armies of specialists that manually enter loan application data and calculate critical metrics to evaluate credit worthiness. Checks for loan stacking are performed, details confirmed, and patterns identified. During this “Pre-Underwriting” phase of a loan, all the necessary data and metrics are gathered for a credit decisioning engine (whether human or machine), which will generate a price or rejection.

ValueStream Ventures’ latest investment, Ocrolus, began with a simple mission: to enable the more time- and cost-efficient collection and analysis of paper documents. Ocrolus allows users to capture loan application documents—many of which are submitted via PDF, scans, or camera images—and efficiently convert that data into a structured, machine-readable format. By combining deep learning technology, optical character recognition (OCR), and a distributed human review network, the team has achieved an incredible accuracy rate of 99.9% at a cost far lower than that of traditional, manual data entry.

Our Ocrolus is driven in part by our thesis that the marriage of Artificial Intelligence and human networks is the best path forward for industries with high accuracy requirements, like Financial Services and Healthcare.

Since its founding, the Ocrolous team has moved beyond raw data, developing analytics that feed into lenders’ subsequent decisioning models and free up lenders to focus their time and efforts on the analyses that best differentiate their underwriting approach.

We’re excited to work with Ocrolus and our fellow investors Bullpen Capital, Laconia Capital, and QED to incorporate new document types and analytics into the Ocrolus offering and to further establish the concept of Pre-Underwriting as a Service. For any of our readers that consume paper as part of a lending process, please reach out to learn more or connect with the team.