Search for a command to run...
Intelligent and autonomous systems are essential for ensuring effective data dissemination in disaster management, particularly for coastal communities and fishermen. Autonomous systems that can make decisions independently are particularly valuable in responding to the dynamic and rapidly changing real-life situations of fishers as they are more prone to natural disasters during their fishing trips. The recent advancements in Large Language Models (LLMs) have made it possible to envision human-like cognitive systems capable of reasoning. In this paper, the authors focus on designing an autonomous AI system, which is capable of alerting fishermen on various imminent dangers in their preferred language. The system is able to fetch information from multiple sources, analyze the collected data, and make alert decisions based on its reasoning capabilities. The system follows a client-server architecture with LLM-enabled agents and the Model Context Protocol (MCP) which allows the LLM to access multiple tools and API to perform its assigned task. The system prototype is implemented using web scraping, prompting, and multi-channel processing, which automate information retrieval and intelligent analysis of existing warning messages from government websites, and generate customized alerts that are both contextual and user-friendly. The system is designed to operate hands-free and receive emergency alerts as voice alerts. The prototype implementation shows promising results, as the system can easily categorize and send multilingual safety voice alerts.