Energy management systems can avoid energy waste, increase savings on the energy bill, increase energy efficiency in the building, and optimize the usage of renewable energy sources to increase the building's sustainability. One of the great challenges, and opportunities, of energy management systems is the human-in-the-loop interaction, enabling users to monitor, control, and operate the facilities as part of the system. This paper proposes a novel human-in-the-loop solution for electric water heater optimization supported by artificial neural networks to predict energy consumptions based on daily and monthly data reading. The proposed solution uses as an application environment a Smart Home that with the use of the Internet of Things establishes the connection between the devices, the use of machine learning to predict the activation of the electric water heater and an Intelligent Virtual Assistant (Alexa) for human-in-the-loop interaction. The case study demonstrates the potential of the proposed solution, as it achieved 47.50% of energy savings compared to the averages of energy consumption between two weeks, considering a real residential implementation. In addition, the average water temperature of Electric Water Heaters during the week with the proposed solution was stable, thus not compromising the user's comfort.