This paper describes work in progress to automate (where possible) computer support for colocated, synchronous, small team meetings. Researchers have made progress providing support for groups in different places and times but there are few face to face support applications. We claim that the behaviours in human interactions are too hard to predict and, consequently, using an algorithmic approach for designing support has been neglected. Current approaches are too prescriptive with their support for “declared” processes when humans are so variable in the way they carry out their duties or their “enacted” processes. We propose that the low level human actions and behaviours must be observed, identified and supported dynamically from a toolset. In this manner, computer interventions can be provided in real time. The main contribution of this paper is an extensible framework called RES, which helps researchers to understand the problem before considering a solution. The framework is informed by established theories of human behaviour. Concepts within these theories are paired with observed behaviours and these pairs tell small stories about the interaction. Aggregating small stories may then describe observable activities in a meeting.