The Neat Company ups its machine learning game

The Neat Company’s machine learning engine, which automatically extracts accounting-related data from small business documents, now supports automated document classification for receipts, invoices and bills of lading.

Neat previously was known for its handheld scanners that accompanied its document-scanning capabilities. Last year, the company discontinued its hardware scanners completely in favor of developing better software that could capture scans from smartphones.

The Neat Company’s software organizes all of a business’s information in one place on the cloud, automatically classifies the document and extracts key information, and then automates the data entry into a company’s accounting system. Because the information is stored online, businesses can share their information with their accountant in real time.

Neat also offers human verification on receipts as they are entered into the system. However, because of Neat’s machine-learning capabilities, the application permits the user to turn this feature on or off.

“We have a unique industry advantage that comes from processing hundreds of millions of receipts, invoices and other documents. Over time, this volume increased the overall accuracy of our automated data extraction,” stated Andrew Schaps, vice president of product development at Neat. “[Optical character recognition] accuracy within the app is so high, users can choose whether or not they require human verification, which they can limit to specific situations like capturing handwritten tips from receipts.”

Also this year, Neat introduced the Neat-Freak Partner Program, which allows accountants and bookkeepers to partner with the company for reseller-related benefits.