Managing coastal emergencies is becoming harder as storms, floods, and other hazards become more frequent and complex. We propose a new approach using artificial intelligence to help make better decisions during emergencies. In our model, an AI agent 'learns' how to respond by simulating different emergency scenarios along the coast. It sees information like wave height, storm surge, rainfall, and wind, along with local details such as where houses are and how close people are to danger. The agent chooses from a range possible actions, such as sandbagging or targeted evacuations, and improves its decisions over time. Early results show the AI learns to respond to the evolving situation focusing on high risk areas, minimizing damage. The system is being trained in a simplified virtual environment but will eventually use real data and connect to more complex digital models. By combining this with other AI tools and working closely with emergency managers, we aim to create a smarter, more reliable early warning and response system for coastal communities.