Advancing Automotive Production: An LLM-based Impact Analysis for Software Updates

In vehicle production, updates of software components, both inside and outside vehicles, can lead to impacts that influence production stability or even halt production. For example, impacts could be that production steps are extended in time. To identify impacts early in the software development process, this paper presents a novel AI-Driven approach for predicting impacts. Therefore, documentation artifacts and incident reports are preprocessed and used with a Retrieval-Augmented Generation (RAG) system to integrate the extracted knowledge into a Large Language Model (LLM). This approach supports detecting and resolving impacts that might otherwise only become apparent during testing or in production.