Reconciling crop productivity and biodiversity maintenance is one of the main challenges of agriculture worldwide. Weed management is recognized to be a key point for ecological intensification in agriculture because weeds can generate severe yield losses but also represent the base of agricultural trophic networks. Research in weed science has often opposed two different perceptions of weeds. Low within-field weed diversity and abundance has either been considered as a sign of efficient weed management or an erosion of the agroecosystem services provided by weeds. However, a recent study in grain-based systems in France highlighted the potential benefits of weed diversity in mitigating crop yield losses. Major yield losses may simply arise from the dominance of a few competitive species. A higher diversity of traits (characteristics) within the weed community should induce complementarity in resource use (light, water, nitrogen etc.) and alleviate weed:crop competition. Thus, weed scientists should try to confirm this relationship in different production situations (e.g. floristic contexts, pedoclimates, cropping systems) and then identify cropping systems which promote weed evenness, either from a taxonomic or functional point of view. Weeding operations should exclusively target competitive and dominant species. However, current weed control practices do not allow to target a specific species in a complex community. Therefore, future studies need to identify if weed diversity could rather be indirectly promoted by diversifying weed management tools, which ought to limit weed density/biomass. The CA-SYS platform (INRAE, Dijon, France) is a unique site in Europe to experiment biodiversity-based forms of agriculture, including a diversity of weed management strategies. The overarching objective of the CA-SYS platform is to design and test the feasibility and performances of pesticide-free agriculture which resorts to (cropped and wild) biodiversity in support of production.