Computer programs are one of the most important tools we use in our daily work. Depending on the profession and area of responsibility, we often use programs that almost everyone knows and uses. However, we also frequently use software products that have been specifically developed or adapted to meet the unique needs of our tasks or to reflect the individual business processes of an organization. In cultural institutions, for example, this includes the planning of performances and rehearsals, orchestra planning, event budgeting, contract management, procurement, staff scheduling, time tracking, the calculation of tariff surcharges, or contact management. The more complex and customized these solutions are, the harder it becomes for users to learn how to work with the product. And the smaller the user group, the less helpful it is to search on Google. Therefore, conveying know-how about highly specialized and customized software solutions is an essential challenge for organizations and is becoming increasingly important as our administrative processes continue to digitalize. At the same time, this training task often hits its limits and consumes significant resources. That’s why we are researching how AI can support this process.

Software manuals are a central and always-available resource for information. Ideally, they describe the software and its functions. However, when a problem arises, the solution is often not that simple.Writing manuals is a time-consuming process. When a new feature is developed, it must be documented. When a feature changes, the description needs to be updated. To help users find their way, a structure and language must be created that bridges the gap between the user’s perspective and the program’s logic. Traditionally, manuals have been written with a “standard user” in mind, but this approach often falls short. We learn more easily when we are addressed individually.