Google's India-Developed Agricultural AI Models Gain Global Traction

Key facts

• Google's agricultural AI models, Agricultural Landscape Understanding (ALU) and Agricultural Monitoring & Event Detection (AMED), were developed by Google DeepMind to analyse agricultural landscapes using satellite imagery.
• Initially focused on India, the models are now being tested by partners in 11 countries across Asia-Pacific and Africa.
• The models are available to the broader ecosystem through APIs and as a data layer on Google Earth, with ALU becoming one of the most popular layers globally.
• CarbonFarm is using the ALU API alongside Gemini to automate field-level insights, including mapping individual field boundaries, for sustainable rice cultivation, targeting 2 million hectares by 2030.
• Terrastack has built a spatial intelligence platform using ALU and AMED APIs, mapping more than 140 million hectares of farmland for agricultural finance.
• In Telangana, the ADeX platform is using ALU and AMED to build digital services for over 5 million farmers, including the Krishivaas app that provides localised information on crop stress, weather patterns, and pest outbreaks.
• Karnataka's Water Resources Department is combining ALU and AMED with local weather and remote sensing data to improve monitoring across about 2.6 million hectares of irrigated land.
• Google is working with the UN Food and Agriculture Organization (FAO) on geoAI4stats, an initiative to incorporate ALU and AMED into the FAO's global CROPGRIDS data repository.
• Alok Talekar, lead for agriculture and sustainability research at Google DeepMind, said the wider adoption of the models is an extension of the team's work on targeted agricultural applications.