Energy demand flexibility from commercial buildings can play a critical role in the ongoing energy transition. There is an urgent need to redirect more research and deployment efforts toward real-world experimentation. Buildings-sector roadmaps overwhelmingly rely on simulations that imperfectly capture reality. We draw lessons from a review of two decades of literature on real-world flexibility and demand response experiments and from our “Living Laboratory” experiences at three major academic institutions in the United States. While the prevailing method is “model first, experiment second,” there is also strong value in “experiment first, model second” and in improving our understanding of a system through experimentation while modeling it. Commercial building clusters on university and corporate campuses offer valuable and often untapped potential. They are both ideal test beds for research on energy flexibility and a significant source of flexibility. Our research agenda provides practical recommendations for conducting and scaling experimentation in these test beds and leveraging experimental findings to improve modeling.