This paper investigates how the management of groups that communicate electronically, such as group formation, can be simplified based on users' context and social relations. This work builds on a framework for Aggregated Social Graphs, where each node represents the relational strength to other users. The strength of a relation is calculated by utilizing information on how we communicate using mobile phone calls, emails, and social networks in combination with additional sources of information such as from calendars. A contextual group management schema is presented where contextual parameters such as tags, locations and objects are used to prune an aggregated social graph in order to automatically form a group. The schema is implemented in a runtime environment based on the Distributed Shared Memory service available at Ericsson Labs. The feasibility of the proposed schema is then studied through a prototype implementation both in a web-browser and as a mobile app. The study shows that a group can be formed automatically and that a lightweight communication session then can be initiated for that group.