Seidler, Marvin

Diploma Thesis:
2114

Topic:
Design and development of an AI agent to provide operational support to a construction company

Editor:
Marvin Seidler

University Professor Responsible:
Univ.-Prof. Dr.-Ing. Dipl.-Wirt.-Ing. Jens Otto

Supervisor:
University Supervisor: Dipl.-Ing. Veit Klopfer 
Company supervisors: Dipl.-Ing. Dirk Peuß (Eiffage Infra-Ost GmbH)

Editing Period:
22.06.2026 until 26.10.2026

Abstract: 
This dissertation deals with the design, development and
economic evaluation of an AI agent based on Retrieval-Augmented Generation (RAG)
for the operational support of a construction company. The starting point
is the observation that established digitalisation tools improve the exchange of structured data, but do not enable the semantic indexing of unstructured company documents. The AI agent, implemented on Microsoft Azure, makes such documents accessible across various business-relevant subject areas by means of natural language queries. The benefits of the system are empirically examined using several real-world use cases and subsequently subjected to an economic evaluation. The thesis concludes with a recommendation for the implementation of the system, as well as a critical assessment of legal and technical limitations.