Feasibility of using an AI-based conversational assistant to support internal technical support staff working with medical devices

Sammanfattning

This thesis investigates the possibility of using an AI-based chatbot as a support tool for technical assistance related to medical devices. The aim is to improve access to relevant technical documentation for biomedical engineers working with troubleshooting and handling of medical equipment.The system is based on document-driven information retrieval combined with language model–based text generation. By applying a retrieval-augmented generation architecture, relevant document segments are retrieved from a knowledge base and used as context for generating responses to user queries.In this work, a proof-of-concept solution is designed and implemented that enables interaction with technical documentation through natural language. The prototype is evaluated with respect to functionality, reliability, and limitations within a technical support context. The results aim to provide an initial indication of how AI-based systems can support access to technical documentation in medical device environments.