This article presents an approach of an assistance system for manual mounting tasks. For this, the three main requirements usability, flexibility and reliability are identified. A mounting assistance system is implemented in a test scenario, with consideration to the previously established requirements. It consists of an industrial laser guidance system that is controlled by a software using machine learning based image recognition. The software determines the status of the actual mounting progress and the location of product components. It controls the laser projection system, thus eliminating the need for any user interaction. Additional instructions are projected onto the work surface to increase the system usability. The results of a structural user test evaluating the mounting efficiency, effectiveness and perceived usability are presented. Finally, the article gives an outlook to further developments.