This project presents a multilingual system designed to analyze stakeholder feedback and classify messages according to sentiment, topic, and urgency. The system was created to support organizations that handle frequent communication involving concerns, safety issues, service requests, and general observations. By combining transformer based language models with rule-based reasoning, the system aims to interpret feedback in a way that feels both accurate and practical in real operational environments. Overall, the project demonstrates how modern natural language processing techniques can contribute to safer, more informed, and more responsive decision-making within complex industrial settings. © 2025 IEEE.