We started our work with Zudello with a state-of-the-art evaluation that compared what was being offered by commercial and Open Source solutions with what was currently in place in Zudello’s pipeline. We concluded that the fast-paced growth in the demand for documents and analytics by Zudello customers required technology that was far beyond what was offered in the market. 



We started with a fast and iterative process, where we focused on addressing the largest bottlenecks in the system, trusting the expert annotators on Zudello’s team to produce the data to refine the model. 



As time passed, our models continued to evolve, reducing the amount of human work required, allowing us to spend more time on other advanced features and, more importantly, allowing Zudello to process thousands of documents per day.