Rising CO2 emissions are contributing to global temperature increase. Renewable energy production promises the dream of sustainable consumption, but if it is not matched with a suitable control mechanism for the electric grid, this goal will be impossible to reach. Towards this, we are studying Fast Demand Response and working on two projects about automated electric load classification and control for the residential and commercial sector. Firstly, we are developing near real time classification of high-power consumption devices. A cheap digital dimmer circuit is used to alter the input voltage signal for several milliseconds. The output signal is analyzed to quickly distinguish load types. Our initial results show that kW-order power savings can be achieved during evening peak demand. Secondly, we are developing presence detection using a Wi-Fi router’s power consumption to quickly and easily predict the number of people in a room. Using this novel approach, a self-reliance power management system which can control power consumption depending on the number of people can be created. We are implementing prototypes and running real-world experiments in both projects.