The process of requirements specification usually begins with understanding the environment, which a designed system will work in. A main source of knowledge about the environment is often given in a form of textual descriptions. Processing such texts by analysts is liable to errors and misinterpretations. The article presents an outline of a tool assisting the process of conceptualisation of reality through automatic analysis of a textual description with the use of techniques of natural language processing. The tool identifies errors and ambiguities in the text, creates a model of its meaning and, on such basis, generates a business model consisting of UML use-case diagrams and class diagrams.