Conversational Demand Response: Bidirectional Aggregator-Prosumer Coordination through Agentic AI

A research team from TU Wien and the Austrian Institute of Technology (AIT) has published a groundbreaking paper on arXiv titled 'Conversational Demand Response: Bidirectional Aggregator-Prosumer Coordination through Agentic AI.' The work introduces a new framework called Conversational Demand Response (CDR), designed to solve a critical bottleneck in smart grids: the lack of transparent, two-way communication between utility aggregators and residential energy prosumers (homes that both consume and produce power).

CDR replaces traditional one-way dispatch signals with a multi-agent AI architecture. In this system, an 'aggregator agent' sends flexibility requests in natural language to a 'prosumer agent' embedded in a Home Energy Management System (HEMS). The prosumer agent can then assess the request's feasibility and cost by calling an optimization tool, and respond with a counter-proposal or acceptance. Crucially, the system also allows prosumers to initiate conversations upstream, communicating changes in their preferences or constraints directly to the grid operator.

The proof-of-concept evaluation demonstrated that these AI-mediated negotiations can be completed in under 12 seconds, making the approach highly scalable. The researchers argue this bridges a fundamental gap, providing the automation and speed needed for grid stability while preserving the transparency and user agency essential for long-term prosumer engagement. To foster rapid innovation, the team has made a significant contribution by releasing the entire system—including all agent prompts, orchestration logic, and simulation interfaces—as open-source code.