Integrating an AI Copilot into a Knowledge Management System forCivil Protection: A Development-Oriented Case Study from Romania

This paper presents the conceptual framework and ongoing development of a local AI Copilot system designed to support knowledge management and decision-making within the Romanian civil protection authority.Anchored in the existing research on AI-enhanced Knowledge Management Systems (KMS), the solution combines semantic indexing, legal version tracking, and multilingual document retrieval to automate the generation of official institutional outputs.Developed as a secure, offline system, the Copilot performs context-aware searches across legislative, operational, and policy documents, and produces structured responses aligned with national and international frameworks.Methodologically, the study applies a design-based research model and illustrates the system's capacity to improve organizational responsiveness and institutional memory.Although not yet operational, this case contributes to the emerging body of work at the intersection of AI, knowledge governance, and public sector performance.