Facing the challenge of the ecological transition of agriculture, biodiversity opens new avenues to enhance ecological interactions and reduce chemical input dependency. Designing biodiversity-based agrosystems requires an agroecological approach that combines key principles: exploring a wide range of concepts and solutions, adopting systemic reasoning, implementing a sitespecific approach, developing an action-oriented process, and maintaining a continuous improvement dynamic. This type of approach has never been developed to harness mycorrhizal fungi, which are key components of soil biodiversity, because their beneficial action on crops depends on complex and underexploited ecological interactions. At present, mycorrhizae are mainly used through industrial inoculants that fit within the productionist paradigm. To shift toward agroecological approaches, we implemented a methodological framework conceived to better address the design of mycorrhiza-friendly cropping systems by sharing knowledge with farmers in four different study areas (Provence, French Guiana, Guadeloupe, and Martinique). This framework includes participative workshops, a board game, and prospective exercises to collect farmers' proposals and the factors that prevent from implementing mycorrhiza-friendly cropping systems. We showed that 90% of the farmers proposed alternatives to industrial inoculants, 50% of them adopted systemic reasoning by combining these alternative proposals. Most farmers understood that they were all potential "mycorrhizae producers". We showed, for the first time through on-farm experiments that valorization of indigenous mycorrhizal fungi strains using a donor plant is an effective practice to increase root colonization before planting (up to a frequency of 95% and an intensity of 32%). Considering the increasing supply of mycorrhizal inoculants and despite the uncertainty of related knowledge, we codesigned innovative practices. Learning communities (technical advisors, high school teachers, etc.) assumed responsibility for continuous improvement in knowledge and practices. Finally, beyond the issue of mycorrhizae, we showed that an agroecological approach could bring stakeholders one step further into the design of biodiversity-based agrosystems.