AutoB2G harnesses large language models to automate building‑grid co‑simulation, eliminating manual coding and enabling real‑time policy evaluation. By integrating reinforcement learning with natural‑language task descriptions, the framework streamlines experimentation across global energy networks.
The Signal
The system builds on CityLearn V2, extending its capabilities to evaluate grid‑level impacts, thereby accelerating research into sustainable, AI‑enhanced energy management worldwide.