Rapid-fire pitch: Vector digitizes documentation of all shapes and sizes, even if crumpled and torn

Photo: Jim Allen/FreightWaves

Photo: Jim Allen/FreightWaves

Technology improvements have increased relevance in most freight-related operations. However, many processes are still paper-intensive. This is a problem for companies that seek to analyze operational data, because extracting data from paperwork is an arduous exercise in itself.

Vector, a San Francisco-based startup, is addressing this with the help of machine learning and computer vision algorithms that can read through documentation and make sense of the numbers as a human would. The company’s lynchpin solution is called LoadDocs, a tool that helps firms improve workflows by collecting information from trucks on the road, digitizing it, and forwarding it to shippers for invoicing.

“We started from humble beginnings. One of my good family friends was being groomed to take over the family business, which was a flatbed trucking company in Oroville, California. One of the biggest risks to his business was cash flow, because it takes a long time for him to get paperwork back from the drivers and create an invoice to send out to his shippers,” said Will Chu, CEO and co-founder at Vector.

“I spent about nine months in Oroville, learning about the business and the challenges he had as a flatbed trucking company. What I learned was that an inordinate amount of time was being spent on manual data entry. It took the company a long time to punch this information back into the accounting system –  both the transportation management system and the enterprise resource planning system, while costing a lot of money.”

Chu realized that automating the data collection part of operations is a key to improving efficiency and thereby fleet margins. At MarketWaves18, Chu explained how automation could help capture data points from documents like the bill of lading and proof of delivery, which if not digitized, would remain in a tabular format and possibly get lost over time. Apart from printed documents, Vector’s machine learning software can also recognize handwriting, which in essence, would help extract and store every ‘scribble’ in a document.

Don’t miss it.  Register today .

Don’t miss it. Register today.

Vector’s user interface is easy to manage – carefully designed to make document submission simple for truckers on the move. The software is trained over millions of data points, and therefore it is quick to recognize and validate documents that might not be in great condition (for example, a document that has been creased or torn across its corners).

However, it was not always this easy, claimed Chu. “When we first deployed our solution, it didn’t go so well. A major lesson was that we didn’t appreciate the environment from which these documents were being submitted. Quite often the carbon copies don’t do well on fax machines, drivers may text raw photographs back to the back office, and the paperwork is ripped and damaged. Sometimes drivers don’t take pictures straight,” he said

This led Vector back to the drawing board, and then the company came out with an upgraded solution that can digitize 1,000 documents per minute.

“I think we all see value in the digitization of the supply chain, but before that could happen, we have to solve a fundamental problem. Data is exchanged and socialized in this industry on paper, as it is the easiest thing to communicate with. It’s the de facto API of the logistics industry. Vector hopes to bring in further digitization by extracting information that is offline and moving it online,” said Chu.