In this article, we build on an on-going participatory research conducted in Corsica to report the use of a heuristic cropping system model as a mediating tool to assist researchers and actors in the understanding of a challenging scientific problem: The agronomical determinism of the acidity of Corsican Clementine, acidulous taste being a strategic typicity attribute for this terroir product. The collaborative research included the participative conception and refinement of an acidity model, the use of the model to design an empirical validation method called Regional Agronomical Diagnosis (RAD), and the up-coming use of RAD’s first results to refine the model. Iterations between actors’ knowledge, model, and farm survey may continue in the 2 following Clementine campaigns, leading to a step-by-step convergence between model and reality. Our acidity model opens up perspectives for Corsican stakeholders, since it could be used in further participatory research to design innovative cropping systems in adaptation to emerging challenges such as climate change, or new varieties. However, the use of a model to understand an agronomical variable or to orientate field observation is not new. What is new is the fact of using such model as an intermediary objet in a participative device. Through its intrinsic infrastructure, the model structured interactions between actors, researchers, and experimental device, and it enabled convergence of representations between opposed epistemological postures. If the model appeared to be predictive, it would suggest that in localized agri food systems, local knowledge are crucial resources which can be channeled by researchers by using intermediary objects as mediating tools in the perspective of addressing complex scientific problems.