Applications on a future Internet of Things require the provisioning of current, relevant and accurate context information to end-points. Context information existing globally require organization into object-oriented models available locally in APIs as current, relevant and accurate views. Moreover, such applications require support for the highly dynamic interactions influencing continual changes in global context information. Existing approaches, such as the web services, are unable to provide this support partly due to the presupposed existence of a network service brokering context information, relying on DNS; or adopting a presence model for context which does not adequately scale. To this end, we propose a distributedframework for the interconnection of end-points and co-located agent entities, whereby agents are provided with local views of a relevant subset of global context information. We show how to achieve relevant local current views of global context information via ranking in an object-oriented context model. The distributed approach realizes the provisioning of context information in real-time, i.e., with predictable time bounds. Finally, we demonstrate the feasibility of the approach in a prototype based on P-Grid. © 2012 ICST Institute for Computer Science, Social Informatics and Telecommunications Engineering.