Rossum provides a solution to automate data extraction from documents with Artificial Intelligence to create a world without manual data entry. Unlike traditional OCR tools, the robot requires no specific rule or template setup — it is ready to extract data from a wide variety of invoices right away. This is thanks to our proprietary deep learning technology that infers the underlying general structure of invoices. Rossum customers achieved field level accuracy of 98%. Customers (or robots) submit their invoices by email or by API, and the extracted fields are returned typically within 30 to 60 seconds. A user interface is provided for the user to validate, correct and amend the data. The inputs are invoice images in pdf or jpg format. The output is in csv, xml or json formats that can be used for direct integration into ERP/accounting system, or a robot can use the data for invoice routing and posting.