Taming the Invoice Review Monster

Abstract

Cost overruns and administrative complexity represent persistent challenges in large-scale infrastruc ture megaprojects. This master’s thesis investigates the potential of using Artificial Intelligence (AI) to identify cost deviations within the invoice review process, focusing on the Swedish Transport Ad ministration (Trafikverket) and its West Link project. Currently, manual review processes suffer from a critical "capability gap", where the overwhelming volume of documentation forces a sampling based approach that fails to detect cost deviations. Through a qualitative case study involving semi-structured interviews with project controllers and managers, this research identifies the bottlenecks created by informational asymmetries between com missioning organizations and dominant contractors. The findings suggest that AI, specifically through automated data processing and pattern recognition, can bridge this gap, allowing project owners to transition toward the role of a "strong owner". The thesis proposes a framework for integrating AI into existing workflows to enhance the invoice review function in projects with cost-plus contracts. By shifting from manual sampling to automated, comprehensive verification, organizations can better manage the administrative complexities inherent in large infrastructure projects.