In the community workshop organized yesterday May 3rd by Microsoft Research, BEYOND colleagues presented their recent work relating to sustainable agriculture. The work uses Earth Observation data and agricultural information to come up with local-specific insights on the performance of selected agricultural management practices, such as crop rotation or diversification. It utilizes recent methods at the intersection of machine learning and causal inference. Its results can be provided to policy makers and other stakeholders for sustainable agricultural planning in light of climate change and increased demand for agricultural products.