UrbanInsight introduces an innovative digital twin platform that leverages edge computing to transform raw data streams from diverse urban sensors—including IoT devices, CCTV cameras, and other data sources—into structured knowledge graph ontologies in real-time. This approach significantly minimizes large-scale data storage requirements and reduces processing times during simulation queries within digital twin environments. The platform utilizes a customized, fine-tuned AI agent powered by a large language model (LLM) to dynamically determine the rules and criteria for data conversion and integration into the knowledge graph. By processing data at the sensor edge, UrbanInsight efficiently prepares time-series, image, and video data for immediate semantic integration, facilitating rapid and accurate scenario exploration and decision-making for enhanced urban