One third of the world-wide energy use and its respective emissions are linked to commercial and residential buildings. Despite this prominent position, buildings are still a passive player in modern energy networks. The industrial and transportation sector are increasingly embedded in an active manner, while buildings still act as unidirectional endpoints and are treated as "black box". Active members of smart grids can contribute to the overall optimization of the energy system by being operated flexibly and by sharing information with the grid. Buildings host a number of significant energy-consuming processes, like heating, ventilation, air-conditioning (HVAC) and lighting. Many processes have operational bandwidths in terms of set-points and scheduling which can be used if needed. Aggregating a number of buildings would lead to even larger flexibility and larger loads that can be dispatched. Strategies like "demand response" (DR; loads, reacting on events in the energy grid) are in its infancy because two key factors are still unsolved: The Smart Grid does not know the states of the load processes, and even if, there is no standard way to communicate them. Both is needed for intelligent algorithms that harmonize loads with grid operations. This is the reason why DR is still open-loop control, where DR-events are broadcast to the loads without knowing the potential consequences. No planning and anticipative reactions are possible in such a system. An intelligent system would take the process state of the customer facilities into account, and would get feedback about the reactions. A traditional DR system can neither estimate the magnitude of a reaction to a DR-event, nor how long this reaction may last, because the loads do not expose information on their current state. It is the goal of the project to close that gap and to investigate in a series of experiments where the limits of intelligent buildings in a Smart Grid are. For this a number of generic load models for buildings must be developed and embedded into an interoperable communication infrastructure. Particular insight is expected by putting building control and grid control into relation, currently these two systems are optimized separately. The investigated objects will be medium and large-size residential and commercial buildings, the test cases will be conducted semi-automatically. Results are figures about the operational potential of "active" buildings and communicable and aggregatable load models, constituting a stepping stone to the intelligent, smart-grid enabled building.