Manual document annotation is one of the most time-consuming steps in both qualitative and quantitative analysis. This thesis describes the design and implementation of a component of a web application called semANT. Specifically, the implemented component consists of a document viewer and an annotation editor with AI-powered features. The editor allows for coding, which is the process of manually marking passages in the text and assigning labels (codes, tags) to them. Two key functions utilizing AI methods have been implemented. The first is automatic annotation suggestion using LLM, where the system suggests suitable passages to the user and assigns relevant tags to them. The second feature is the suggestion of the most suitable tags for a manually marked section of text using zero-shot classification with the mDeBERTa model, which evaluates the most relevant tags and then displays them in the user interface. The logic of these features is separated into a standalone Python library called Topicer. User testing showed a reduction in coding time when using automatic suggestions. Testing the agreement between manual coding and the classifier achieved an agreement rate of 56—85%. The results suggest that integrating these features into the annotation editor has the potential to reduce the time required for coding in qualitative analysis.