Innovation has become crucial in the strategic orientations of agricultural research institutes, and is declared as critical for future agriculture to meet the challenges raised by societal expectations. This emphasis on innovation requires to orientate agronomic research to design or design-support studies (Klerkx et al., 2012; Prost et al., 2016; Salembier et al., 2018), which contrasts with a persisting "academic polarization" (Bonneuil and Thomas, 2009) and a widespread view of innovation as resulting from the aggregation of available scientific knowledge. In these design studies, agricultural researchers do not aim primarily at producing knowledge that describes the world as it is, but rather at designing tools, methods and processes that support the actions of farmers or their advisors to achieve desired farming systems. Moreover, Design Studies show the existence of close links between design and knowledge production, (Eekels and Roozenburg, 1991; Hatchuel and Weil, 2009; Owen, 1998). In agronomy, various studies question the knowledge that supports these design processes (Berthet et al., 2015; Duru et al., 2015; Toffolini et al., 2017), but little consideration is given to the scientific original knowledge that is generated during design processes. Based on a cross-analysis of empirical cases in which agricultural researchers have designed innovations with a diversity of actors from agricultural sectors, we analyzed the production of scientific knowledge that takes place during design processes.