The increasing complexity and volume of quality data in modern manufacturing environments, driven by the adoption of 3D optical scanning technologies, has outpaced the capacity of human inspectors to interpret such data consistently and efficiently. This paper presents an AI-assisted methodology meant to enhance the interpretation of quality inspection results, like deviation reports generated from CAD-vs-Mesh analyses. The proposed approach translates raw inspection outputs into context-rich, semantically meaningful narratives. This multi-agent framework integrates both large language models (LLMs) and vision-language systems. Multiple AI agents in the system perform specialized roles. After the initial data parsing and deviations are identified and interpreted by two independent vision agents. Their output is then analyzed by a language model which returns a consolidated inspection report that can then be reviewed and validated by a human expert. A demonstrator system is implemented on an automotive component to showcase the framework’s capability to generate detailed and actionable interpretations from complex 3D inspection data. Results indicate that the system has the potential to enhance efficiency and interpretative depth compared to traditional reporting methods. The study also identifies current limitations in vision model spatial awareness and image resolution handling, and outlines directions for future research, including model fine-tuning. This work contributes to the human-centric vision of Industry 5.0 through the augmentation, rather than replacement of expert judgment in quality assurance workflows.