The team developed a disaster resilience platform for Connecticut that centralizes emergency preparedness information, evacuation guidance, and shelter routing into a single, interactive system. The platform uses an agentic AI workflow powered by a large language model (LLM) to interpret natural language queries and dynamically retrieve relevant information while generating optimal routes to nearby emergency shelters. A routing agent integrates with external mapping services to compute distances, directions, and navigation steps, with additional geospatial analysis to assess potential risks such as flood-prone areas. This system demonstrates how AI and geospatial technologies can provide accessible, real-time decision support during emergencies.