GLISSA: A GenAI-Augmented Framework for Real-time Situation Awareness of power Infrastructure based on Crowd Sourced Data

Monitoring critical infrastructure (CI) like national power networks in developing countries is hindered by the limited coverage and high cost of traditional data collection methods (e.g., sensors, SCADA), making real-time situational awareness difficult during service disruptions. In contrast, social media platforms like X (formerly Twitter) offer a rich but underutilized source of real-time user-generated data. However, its unstructured, noisy, and diverse nature challenges traditional AI methods. This paper proposes a fully automated pipeline leveraging Generative AI (GenAI), specifically Large Language Models (LLMs), to extract and synthesize actionable insights from social media. The system includes Twitter API-based data ingestion, LLM-driven filtering and classification, geolocation inference, and visualization. A case study in Uganda validates its ability to detect power disruption events, addressing crowd-source data challenges. Keywords: Infrastructure Monitoring, GenAI, LLMs, Social Media Analytics, Real-Time Systems, Uganda