Leveraging AI for Efficient Information Retrieval in Technical Documentation Databases

Abstract

This thesis investigates inefficiencies in retrieving technical documentation at Deltamarin and proposes an AI-based solution to improve the process. A company-wide survey, with a twenty percent response rate, found that employees spend nearly one hour per day locating and understanding documents. Key challenges include inconsistent file naming, fragmented storage and unintuitive search tools. Based on these findings, the thesis develops a two-year roadmap for implementing Microsoft Copilot, selected for its integration with existing systems, data protection, and financial feasibility. The plan includes preparatory work on data and practices, pilot testing, user training, integration with the M-Files system, and final evaluation. Financial analysis shows that even modest gains in efficiency could justify the investment, while risk analysis emphasizes the need for user engagement, proper training, and technical validation. The study concludes that combining AI tools with improved documentation practices can reduce inefficiencies, enhance employee satisfaction, and improve overall operational performance.