Software update releases have become an integral part of automotive final assembly, yet their effects on production stability are difficult to assess because the relevant evidence is scattered across error tickets, workflow notes, patch notes, and quality indicators. This book develops and validates an AI-based method for release impact analysis that integrates these fragmented sources into an auditable relational evidence base, recovers missing links from technical free text with large language models, and lets a constrained agent answer production-oriented impact questions through logged read-only SQL queries. A systematic evaluation of nine language models under industrial offline constraints and an integrated proof of concept with a longitudinal First Time Quality analysis over 47 production weeks show that release impact analysis is best understood as an auditable evidence integration task rather than an isolated prediction problem.

Aiman El Asad earned his Ph.D. as a doctoral candidate at the Institute for Automotive Engineering (IFS) at the University of Stuttgart, in the Department of Automotive Mechatronics. His research focuses on AI-based methods for verifying vehicle software, particularly on the impact analysis of software update releases in vehicle production.