Executive Summary Task 5.4 aims at integrating the previous WP5 tasks to perform actual control actions according to the devised control strategy. The bridge between the words optimal and control are realised by looking at the results received from the optimiser and then implementing real-life control actions over devices and recommending demand response actions for consumers. Task 5.4 performs advanced data analytics for optimal control and includes integrating several tasks from WP5. First, it considers the principles of the REACT user engagement approach. The optimiser provides a day-ahead planning strategy for energy end-users regarding asset utilisation with the possibility of re-running the process in intraday intervals. The REACTforecast-optimise-controlloop represents an arrangement of services that communicate with each other and provide automatic management of the assets (e.g., batteries, heat pumps, etc.) so that the power system can achieve overall benefits. The forecast starts with data acquisition on pilot sites, which is subsequently utilised for training machine learning models for weather and energy predictions. The outputs of these prediction algorithms are then fed into the optimisation engine. The optimisation engine is an integral part of the REACT platform designed to provide setpoint values to ensure the efficient day-to-day operation of the system. It gives recommendations for optimal load modifications and asset management. The outputs are processed by the electric domain battery and inverter control orchestrator, and the thermal domain Building Simulator. The Building Simulator schedules hourly control actions to be performed for the next day. The Consumer Recommender Service can generate and send recommendations in the form of PUSH notifications to the REACT Android mobile app. The control actions are sent to all the connected devices through the REACT cloud platform, including heat pumps and batteries. Task 5.4 developed different battery control strategies. In addition, an Energy Box has been designed to achieve direct load control. The report discusses the OpenMUC implementation in detail (channels, parameters, etc.) of the battery controllers, power-to-hydrogen system (CTS H2home), and the Electric Vehicle bidirectional charger (Wallbox Quasar). The MIDAC and MELCloud implementations are also discussed from the REACT platform integration viewpoint. We also discussed the plausibility of additional control for larger heat pump systems for large buildings. Finally, we test and validate the tasks in two steps. First, tested and validated the control functions from the OpenMUC energy gateway perspective. Then we demonstrated how the envisioned control loop comes into realisation within the overarching umbrella of the WP5.