Beyond Human Performance: Multi-Agent AI for GxP-Compliant Pharma Manufacturing

In high-stakes pharmaceutical manufacturing, visual quality control and automated defect detection are critical for maintaining regulatory compliance and ensuring patient safety. Our recent paper, “Beyond Human Performance: A Vision-Language Multi-Agent Approach for Quality Control in Pharmaceutical Manufacturing”, explores how specialized vision-language multi-agent architectures can collaborate to detect subtle particulate, packaging, and vial defects with unprecedented precision—surpassing traditional baseline computer vision and human inspection thresholds.

By leveraging consensus mechanisms, specialized domain agents, and explainable audit trails, this multi-agent system provides both high accuracy and the rigorous validation required for GxP-compliant manufacturing environments.

Authors: Subhro Jyoti Mandal, Leila Rachidi, Prakhar Jain, Matthieu Duvinage, Sander W. Timmer
Preprint: arXiv:2602.20543
Direct PDF: Download Paper PDF (arXiv)