Diseases in vines, notably Downy Mildew, cause economic losses in the viticulture sector by affecting the oenological quality of grapes in infected vines and causing plants to die. The identification of these diseases usually happens late, soon after the grapevines present damage in leaf physiology, characterized by dryness of the leaves. Although this is a widely known problem, current research is still limited, particularly with regard to the proactive identification of disease incidence. This paper, therefore, proposes VitForecast, an IoT approach to aid in the prediction of diseases in grapevines. VitForecast uses the Internet of Things (IoT) devices and Artificial Intelligence techniques to collect microclimate data and make predictions about the favorability of grapevine contamination. To this end, a disease prediction workflow is proposed, a component-based architecture to support different prediction strategies, and a layered architecture to facilitate understanding and evolution of the approach. VitForecast was implemented through a mobile application and used IoT devices to collect and transmit microclimate data, including temperature and humidity sensors, Raspberry PI, and others. The case study carried out demonstrated the feasibility of the approach, as well as the effectiveness of predicting the favorability of grapevine contamination by Downy Mildew.