The efficient production and handling of multilingual documents of various nature is one of the major challenges to be faced in the short term by international organizations, public services and industry in Europe. The project addresses precisely this challenge by developing a multilingual generation system for texts describing bureaucratic procedures, starting from language independent specifications and making use of advanced natural language processing (NLP and knowledge representation techniques. Three languages are taken into consideration: English, German and Italian. The use of the GIST system by the selected user groups will have an important role in showing the benefits of applied NLP to other users, especially Public Administrations in multilingual areas. The requirements of the GIST system have been determined through a preliminary task-analysis study and interviews with members of the user. The criteria for the generation of good quality target texts and the information to be included in the Domain Model have been determined starting from real texts. Existing modules have been adapted and extended. Specifications for implementation of the text planning component have been provided defining the knowledge sources necessary for the construction of an optimally structured text plan .