AI maps 9,300 Brazilian soy facilities, unlocking supply chain traceability
Powerful new automated image analysis identifies soy facilities with over 90% accuracy - turbocharging transparency around logistical links in supply chains.
Oxford, United Kingdom (03 Mar 2026) – Trase has pioneered a groundbreaking solution using high-resolution satellite imagery and advanced multimodal AI to automate the identification and mapping of agricultural facilities at scale, starting with the vast soy supply chain in Brazil.
Trase has developed a solution that uses two state-of-the-art geospatial foundation models – AlphaEarth and Clay – to automate the identification of agricultural facilities at scale. The technology enables Trase to scan vast agricultural areas to detect the distinct physical signatures of silos and warehouses that were previously unknown.
“Providing a detailed soy facility dataset enables governments, civil society, and retailers to better scrutinize and verify sustainability claims, as well as identify high-risk deforestation areas linked to production”, says Jailson Soares, Trase data scientist.
Rather than waiting for third parties to provide information, Trase’s proactive mapping system helps identify facilities, including those in remote areas. The data shows the most accurate picture to date of where soy is actually being produced, stored and processed before it reaches global markets.
“The hidden middle, a complex network of silos and warehouses where raw commodities are stored and mixed, has long been the weak link in traceability,” explains Helen Belfield, Trase co-director. “Without detailed facility data, companies buying soy cannot know its true origins, as batch mixing masks environmental risks. This innovation is critical for mitigating deforestation and supports government efforts like the EU Deforestation Regulation (EUDR) by enabling effective enforcement and chain-of-custody.”
Trase identifies facility locations and functions using automated visual analysis of high-resolution satellite imagery. Trase uses a sophisticated AI workflow to recognize visual markers like vertical silos, arched warehouses, and temporary silo bags, and confirms soy facilities with over 90% accuracy. This process ensures the resulting dataset is a validated map of the soy supply chain's logistical facilities.
The method also integrates reasoning agents to determine facility ownership by connecting physical assets with corporate entities. It integrates unstructured data from registries, tax databases, and digital maps, then employs advanced fuzzy matching algorithms to synchronize these noisy datasets. This links facilities to owners, even with inconsistent or misspelled names, transforming fragmented web data into a structured database identifying both the facility location and its owner.
Trase’s success in Brazil builds on its existing mapping work for slaughterhouses, cocoa cooperatives, and palm oil mills. The Trase team is now focused on operational intelligence and addressing remaining challenges in automation.
About the method
Jailson Soares, data scientist in the Trase team, used AI tools and techniques to identify the soy silos and the insights referred to in this release, in close collaboration with consultant Vivian Ribeiro. These AI and data insights were overlooked by Harry Biddle, senior data engineer for the Trase platform.
For further information, contact:
Gisele Neuls, Trase Communications Lead at Global Canopy | g.souzaneuls@globalcanopy.org | +44(0)7925 128 159
About Trase
Trase is a data-driven transparency initiative led by Global Canopy and Stockholm Environment Institute (SEI). We map international trade in agricultural commodities and provide open-access data, insights, and tools that help companies, financial institutions, and governments strengthen accountability for their sustainability goals. www.trase.earth
