At Rossum, we are developing a cloud-based data extraction platform that uses deep learning to read and understand semi-structured documents like a human, eliminating the need for rules and templates. Today, we are thrilled to announce that we’ve secured $4.5 million investment to make this dream come true.
We’re on a mission to rid the world of manual data entry once and for all with Rossum. We’re thrilled to announce that aid for our mission has arrived: we’ve secured $4.5 million in a seed round from LocalGlobe, with participation from Seedcamp, Miton, and StartupYard. We’ve also had the good fortune of securing resources and mentorship from these private investors:
- Elad Gil, Twitter’s former VP of Strategy and investor in Airbnb, Square, and Pinterest
- Michael Stoppelman, investor in Wish and Lyft, and the former SVP of Engineering at Yelp
- Vijay Pandurangan, investor and advisor for Wish and Get Room, and former Director of Engineering at Twitter
- Ryan Petersen, founder and CEO Flexport and Import Genius
Automating data entry
Every day throughout the world, an estimated equivalent of 100 human lifetimes are spent on manual data entry - and that’s just for invoices! To help stop this insanity, we focused on building a platform that could streamline the invoice data capture process. Developed in Prague, Czech Republic, Rossum extracts data six times faster than manual entry methods, reducing document processing costs by up to 80%.
Rossum was built by our company's three founders, former AI PhD students Tomas Gogar, Petr Baudis, and Tomas Tunys. Petr’s work is credited in Google’s scientific paper on its historic AlphaGo AI victory in 2016, Forbes named Tomas Gogar one of its 30 Under 30 in 2018, and all three were named among Financial Times' New Europe 100 Challengers in 2017.
Instead of using traditional optical character recognition (OCR) extraction software, our technology is leading a trend known as cognitive data capture, in which computers learn to understand documents similar to the way a human does.
Traditional OCR-based systems rely on specific rules and templates for every format of document they receive, and they read each document character by character. For even a small business, this can require the creation of hundreds, if not thousands, of templates, and every new format requires a new set of rules. This leads to higher costs, complex third-party integrations, and a dependency on developers to maintain these systems.
Cognitive data capture uses deep learning to scan a document like a human would, regardless of the file’s style and formatting. Our software runs in the cloud, eliminating the need to install cumbersome applications, enabling easy integration with existing systems, and offering the option of usage via a web browser.
Designed to help humans, not replace them
We didn’t develop Rossum to replace employees - quite the contrary. Our goal is to provide human operators with a tool that helps them create more value. We want Rossum to enable businesses to be more flexible and reliable for their customers, and help employees focus on tasks that are more complex and require creativity. Rossum’s accuracy rates average at around 95%; when our software is unable to identify data fields, it asks the user to provide the required information. Every time our platform gets this feedback, it remembers it, which makes it smarter and more accurate, eliminating the need for template and rule creation.
Rossum’s growing corporate user base includes several Fortune 500 enterprises and spans every continent. Siemens happens to be one such company. “When we first started using Rossum’s technology, we couldn’t believe how accurate it was,” said Martin Nedopil, a Siemens Data Capture Expert, adding, “It was so impressive, we were convinced its results were being backed up by manual labour when they weren’t. It was almost too good to be true.”
Currently, Rossum primarily helps users process invoices and similar documents, like delivery notes. Our platform can also process documents across many segments including accounting, logistics, insurance, and real estate management. We plan to use this round of investment to further develop our technology for use in multiple sectors, open a US office, and continue our global expansion.
“Technology should make data entry easier and cheaper, but businesses have become too reliant on using old systems that no longer meet their needs. Rossum solves these problems without complicated, clunky integrations; without teams of developers; and without high costs. Our solution is smart enough to be tailored to suit any type of business and it’s scalable to work with even the largest of firms,” explains Rossum’s CEO and Co-Founder Tomas Gogar
LocalGlobe partner Mish Mashkautsan said, “Until now enterprises have depended on data capture solutions that took months to set up and required heavy maintenance. With Rossum they can now deploy in minutes with high accuracy out of the box across countless use-cases. Their product facilitates intuitive feedback which quickly boosts accuracy and efficiency even further. We’re delighted to back Rossum as they build the future of cognitive data capture.”
Angel investor Elad Gil, who has backed the likes of Stripe, Airbnb and Coinbase comments, “Invoice data management is a huge unsolved problem. Rossum’s traction shows that the company is in a great position to solve this problem as well as tackle many other data entry tasks using its highly versatile platform."
Reshma Sohoni from SeedCamp said: “Rossum’s technology is a game-changer for business. We’re excited to work with such a passionate and highly skilled team to bring the cost and time savings of its AI data-extraction tool to even more businesses.”