Scaling Agri Resilience: From India to the World
Google AI models provide field-level insights
The global agrifood sector continues to face significant pressure. While an estimated 2.1 billion people – over 25% of the world’s current population – were food insecure in 2025, global food production must increase by up to 50% to support 9.7 billion people by 2050. The agriculture ecosystem accordingly needs to develop targeted interventions to address present and future food security concerns.
Our AnthroKrishi team built Agricultural Landscape Understanding (ALU) and Agricultural Monitoring & Event Detection (AMED) models to support this effort. Originally capturing India’s landscape, the outcomes of models were shared with trusted testers in Asia-Pacific last year, and have recently started providing agri insights for 6 nations in Africa: Kenya, Uganda, Ghana, Rwanda, Nigeria and Zambia.
While these models’ outputs have been freely accessible as APIs, we’ve also integrated them directly into Google Earth. We are thrilled that the ALU data layer is one of the most popular layers on Google Earth globally today.
ALU layer on Google Earth
Beyond this enthusiasm, we’ve been deeply inspired by the transformative impact that the solutions built with these models have been delivering across the Indian and global agri ecosystem.
Growing sustainability of food production
CarbonFarm's digital monitoring platform (Source: CarbonFarm)
CarbonFarm has used the ALU API and Gemini to automate field-level insights for its digital monitoring platform that advises on farming practices that reduce the environmental impact of rice cultivation. The Google AI-supported insights have not only replaced paper-based reporting, but are also helping strengthen rice farmers’ access to carbon-credit and climate-finance programs.
The platform is currently deployed across 12 countries, and is expected to expand to 20 countries, supporting CarbonFarm’s ambition to support 2 million hectares of low-carbon rice by 2030.
Strengthening farmer access to agri credit
Terrastack has built a spatial intelligence platform on the ALU and AMED APIs to help Indian farmers access formal credit and climate support mechanisms. Using the APIs, the platform has integrated land records, farm-level satellite data, crop activity, climate signals, and market data for over 140 million hectares of farmland, nearly all the land under cultivation across India 1 .
This platform has reduced the need for physical field visits, while enabling Terrastack to generate three key insights with the APIs – land record reconciliation, biofuel procurement, and farm revenue estimates – which can enable India’s agri value chain to make faster and more accurate decisions that benefit farmers.
Supporting India’s Digital Public Infrastructure for agri
Beyond private sector innovation, Google DeepMind’s agri models are also playing a prominent role in the development of Digital Public Infrastructure for agriculture across India.
The Government of Telangana has integrated the ALU and AMED datasets into its Agriculture Data Exchange (ADeX), which is supporting ecosystem collaborations and innovations to benefit the state’s 5 million+ farmers. As part of this, ALU is helping the state’s agri department reconcile past field survey findings, while the nonprofit alliance NaPanta has integrated AMED into specialized applications that provide farmers early warning indicators and instant crop stress diagnosis. ALU and AMED datasets are also supporting a pilot of the state’s Krishivaas application, which generates actionable, hyperlocal advisories on crop stress, crop-specific weather patterns and localized pest outbreaks.
Scaling digital agri infrastructure to the world
Our agri models are now also helping shape digital public infrastructure at a global scale.
The Food and Agriculture Organization's (FAO) CROPGRIDS data platform leverages national agricultural information to improve FAO's capabilities to monitor sustainability in agriculture at planetary scales.
Building on this effort, the new geoAI4stats initiative at FAO plans to integrate the ALU and AMED datasets into CROPGRIDS to strengthen the depth and frequency of agricultural intelligence available to member countries. The new features will help automate crop detection, generate agricultural maps, and enable faster updating of agricultural statistics.
Supported by $2.5 million in funding from Google.org as part of the AI Collaborative: Food Security, the FAO geoAI4stats project aims to help countries leverage foundational AI models and local datasets to investigate and address current planning and monitoring needs in climate resilience, crop insurance, fertilizer subsidies, and sustainability monitoring.
Strengthening water management
The Karnataka Water Resources Information System (KWRIS), developed under the state’s Water Resources Department, has combined ALU and AMED with localized weather and remote sensing data to strengthen dynamic water management across the state’s 2.6 million hectares of irrigated area. Our model outputs are contributing to the state’s river basin planning and localized food security.
A collaborative future for global agri resilience and food security
Securing our global food supply requires ecosystem-wide collaboration. For our part, the AnthroKrishi team remains committed to bringing the advances of AI to enable increasingly effective solutions that promote local agri resilience and a more sustainable future for everyone.
