Mandatory e-invoicing is coming to France, Belgium, and Poland in 2026, and the Czech Republic’s Rossum is ready. Its Aurora AI focuses on transactional documents, such as invoices, packing lists, or sales orders, distinguishing itself from its competitors with a focus on speed and precision that’s crucial for modern businesses.
Founded in 2017 by three PhD students in the AI field, Rossum has come a long way since founders Tomas Gogar, Petr Baudis and Tomas Tunys dropped out of school to start the business after encountering a “recurring business problem.” That problem? “The inefficiencies of document processing. Hours wasted on manual data entry, frantic searches for missing information, endlessly chasing approvals.”
Determining that the problem remained unsolved, they created Rossum with a focus on driving unmatched productivity and accuracy through its AI-first, cloud-native approach. The ultimate goal is to transform business workflows end to end, addressing risks, enhancing business relationships, and unlocking real-time strategic insights from transactional data.
Aurora AI’s core is a proprietary Large Language Model created specifically for transactional documents, trained on one of the largest datasets in the industry, containing millions of documents with detailed annotations. Through its three levels of training, Rossum Aurora achieves human-level accuracy almost instantly, while being designed to provide enterprise-grade safety.
Gogar said Aurora took two years of “meticulous development” before it was unveiled last year. By that point, it had achieved a 10x reduction in the number of documents needed to reach desired accuracy, an average 37.6% error reduction, with zero hallucinated values, and an accelerated discrepancy handling through generative AI custom emails straight into existing workflows.
That’s why when Rossum claims the company “leads the charge in intelligent document processing,” it doesn’t take AI to know it’s telling the truth.