Abstract Energy storage control, load scheduling, and indoor user comfort management are perceived as key management solutions for electric industry in the building sector. Nevertheless, requirement of a-priori knowledge on system inputs (i.e., renewable energy generation process, load arrival process, and dynamic price signals) raises concerns about the ability of existing building energy management solutions to accurately adapt to real-time needs in energy generation, demand, storage, and indoor comfort feel. Conversely, with the consideration of unknown dynamics of system inputs, a one-slot-look-ahead virtual queue stability based Lyapunov optimization technique is employed in this article for a real-time energy and comfort optimization in grid-connected solar integrated smart buildings. The goal is to minimize an average aggregated system cost through a real-time joint optimization of electrical and thermal load scheduling delays, energy procurement cost from controllable generators and external grid, electrical and thermal energy storage degradation, and indoor user comfort feel. It is also shown that the joint optimization problem is separable into subproblems which are sequentially solved to obtain all solutions in closed-forms. The solutions are proved as asymptotically optimal, and can be easily implemented in real-time building energy and comfort management scenarios especially when the statistics of system inputs are unknown and arbitrary. The proposed algorithm is validated through simulations where it is tested in different weather conditions. Results show that the proposed algorithm can achieve an average monthly energy procurement-and-operations cost reduction up to 16.37%, while meeting building’s energy and comfort requirements.