Researchers at the University of Cambridge have developed Tessera, an artificial intelligence foundation model designed to monitor crop types on small, fragmented farms using satellite imagery. The technology targets a persistent gap in remote sensing: conventional tools are built for expansive industrial plots and routinely miss the sub-hectare fields that produce a significant share of global food. By transforming raw satellite feeds into compact numerical embeddings, Tessera captures seasonal land-use patterns rather than depending on isolated images. This methodology effectively overcomes edge-detection limitations, where satellite pixels frequently span multiple crops or boundary features, a factor that traditionally obscures small fields from automated classification. Austrian field trials confirmed that Tessera classifies crop types more accurately than established methods while utilizing only eight percent of the required computational power. The system operates without manual parameter tuning, offering a substantially cheaper and more scalable alternative to legacy monitoring infrastructure. Lead author Madeline Lisaius, a doctoral researcher in Cambridge’s Department of Computer Science and Technology, noted that precision gains at this scale directly impact high-stakes food security decisions. International agencies, including the United Nations Food and Agriculture Organization and the World Bank, rely on satellite crop maps to project yields and calibrate grain import volumes. Even marginal improvements in data reliability can determine whether a country purchases enough surplus to avert domestic shortages. The research, detailed in the ISPRS Annals of Photogrammetry, Remote Sensing and Spatial Information Sciences, will be presented at the ISPRS 2026 conference in Toronto. While the Austrian trials validate the model’s technical performance, the development team emphasizes that full integration into governmental planning frameworks will take several years. Deploying Tessera for actual policy guidance will require cross-referencing AI outputs with on-the-ground agricultural data and accounting for regional socioeconomic variables. Despite this timeline, Lisaius advised policymakers to allocate resources toward the technology now, highlighting its streamlined architecture and low computational overhead as key enablers for rapid scaling. As climate instability and trade disruptions heighten vulnerability in global food systems, Tessera establishes a foundational pathway for more resilient, data-driven crop monitoring across the world’s most fragmented agricultural regions.