Unstructured Metadata Generator Agent

This agent adjusts energy generation forecasts based on real-time and forecasted climate data to support proactive grid balancing. It works autonomously, without human intervention: it evaluates discrepancies between forecast and real-time weather inputs, reasons about their impact, and updates energy models accordingly. This supports other agents in making accurate and timely decisions, such as triggering backup power or anomaly alerts.

The agent reduces energy misforecasting due to weather changes, optimizes backup activation, and enhances the resilience of the grid in dynamic environmental conditions. Its primary users are utility providers in the energy sector. It is built with a Vertex AI Prompt Model (code-interpreter), BigQuery, Pub/Sub, a Playbook, and an optional Dialogflow CX interface.