A method, a computer program product, and a computer system for a spatially aware virtual workout assistant. A server receives from a mobile device a view of a surrounding environment of a user, wherein the view is captured by the mobile device. The server extracts one or more elements from the view. The server recognizes one or more objects suitable for exercising, by using a model for visual data recognition. The server determines one or more exercises that can be performed with the one or more objects, by searching datasets. The server sends to the mobile device data associated with the one or more objects and the one or more exercises. Based on the data associated with the one or more objects and the one or more exercises, the mobile device instructs the user to locate the one or more devices and to perform the one or more exercises.