Generalized legal documents, for which the positions in the text are known for the individual expressions of a contract, can be used firstly to support the approval process of new contracts in an automated way and secondly to provide new legal documents as a contract generator. This paper shows, with the help of known legal texts, how formulaic text sections can be identified and frequent individual expressions can be classified in order to be used as pattern sections. Areas of application are presented and existing potential for legal tech applications is shown.