This report concerns the implementation of an AI-driven assistant for managing, auditing and reviewing document metadata. The idea stems from the inefficiency and major resource demand presented by the current, manual, document auditing process which the AI-driven solution seeks to improve. To research whether the AI-based solution is a viable alternative or not, we created a prototype AI-assistant using the DONUT AI model, which we integrated into a web-based application for automated metadata processing. We allowed the model to analyze various documents which, through statistical analysis, we were able to procure results from. The results showed that there is a lot of room for improvement for our application, especially when evaluating it's performance relative to time. However, the results showed great potential for future research as it performed well relative to the models ability to accurately identify and highlight mistakes in the documents that we're being reviewed. In summary, the application that was developed as a result of this degree project was well received by the external company, Synergetix Consulting AB, which we're the ones who facilitated this degree project.