This paper uses an Activity Theory framework to evaluate the use of a novel, multi-screen, non-linear presentation tool. The Thunder tool allows presenters to manipulate and annotate multiple digital slides and to concurrently display a selection of juxtaposed resources across a wall-sized projection area. Conventional, single screen presentation systems have been frequently criticised in the literature. Yet very little work has investigated the use of multi-screen alternatives, particularly within authentic activity systems. We placed the Thunder tool into an established activity system of student presentations in University-level Architecture education and examined how that system was re-mediated by participants' use of the tool. Our instruments included in-room and video-based observations, questionnaire responses and transcribed interviews. We analyse how using Thunder in this setting re-mediated conventional activity structures in terms of interactivity, efficiency, cost, failure and serendipity. Drawing actions became prominently used to support design arguments, resulting in the "easel" screen becoming a site for struggle between students and tutors. Slide manipulation was not well managed by students and yet the disorganised wall projection area nonetheless supported a directness of communication between participants that was novel for the activity. Students' historical learning costs, in terms of time and money, were removed, serving as the basis for the more fluid, contingent interactions, while participants' perception of tool failure was heavily influenced by prior expectations and historical activity forms. Novel forms of serendipitous discovery were enabled by the juxtaposition and annotation of materials, but such discovery was possible for both student-presenters and their audience and could be used to challenge presenters' narrative. We conclude by discussing how the re-mediation we document is pertinent to the contradictions in the activity system and suggest that, for the particular activity system we studied, a new leading form of the activity system would need to emerge if tools such as Thunder are to better mediate student presentation activity. (Contains 7 figures and 5 tables.)