As problems become increasingly complex, science educators need to better understand how new knowledge is constructed and applied in heterogeneous team collaborations, and how to teach students to productively engage in these processes. We discuss the emergence of insights in collaborative sensemaking and suggest a model that articulates the collaborative nature of this process. The underlying premise that guided the analysis and modeling is that the individual collaborators and the available tools in productive collaborations function as a distributed cognitive system. We illustrate the utility of the model through a fine-grained discourse analysis of a snippet from the work of a group of three students during a lesson for pre-service physics teachers. The group worked on an authentic ill-defined problem that none of them could solve alone. The analysis details the evolvement and systematization of the openly shared knowledge structure and the emergence of local insights. [This paper was published in: "2024 PERC Proceedings," edited by Ryan et al., American Association of Physics Teachers, 2024, pp. 361-366.]