In this paper we report on the initial activities carried out within a collaboration between Consob and Sapienza University. We focus on Information Extraction from documents describing financial instruments. We discuss how we automate this task, via both rule-based and machine learning-based methods and provide our first results.