Remote assistance provides a communication bridge for users engaged in different locations. However, understanding how to design such systems in IoT is a challenging issue given digital representations are not the same as sharing a physical space. In this paper, we present a Remote Assistance Platform (RAP) that is designed to facilitate task guidance between an instructor and one or more remote operators. This includes the support of visual communication using annotation tools that augment information from a live video stream. Two user studies were performed to evaluate co-located and remote interaction. In the first study, dyads interacted with paper-based instructions while situated in the same location. In the second study, different dyads remotely performed the same tasks, assisted by using a smartphone or smart glass display. Overall, our findings found significant differences in communication behaviour based on the type of collaborative environment and information modality used. A short review of these results is discussed.