How Google’s India-First AI Models Are Scaling Agricultural Insights
26 August 2026, New Delhi: The global agrifood system is under growing pressure to produce more food while responding to climate change, resource constraints and the need to support increasingly vulnerable farming communities. An estimated 2.1 billion people, more than a quarter of the world’s population, faced food insecurity in 2025, while global food production may need to increase by up to 50% to feed a projected 9.7 billion people by 2050.
Meeting these challenges will require more than conventional approaches. Better agricultural data, timely insights and targeted interventions are increasingly important for improving productivity, strengthening resilience and supporting more sustainable food production.
Against this backdrop, two artificial intelligence (AI) models developed by Google DeepMind are being used to generate agricultural insights at scale.
Agricultural Landscape Understanding (ALU) and Agricultural Monitoring & Event Detection (AMED) use satellite imagery to map agricultural field boundaries and monitor agricultural activity. Initially developed with India’s agricultural ecosystem in mind, the models are now being tested by trusted partners in 11 countries across Asia Pacific and Africa.
Their outputs have also been made freely available to the wider ecosystem through APIs and Google Earth. The ALU data layer has become one of the most popular agricultural layers on Google Earth globally.
The models are being used by organisations across agriculture, government and development sectors to build applications aimed at addressing challenges ranging from sustainable cultivation and farmer access to credit to water management and agricultural policymaking.
From field-level data to agricultural resilience
One application is helping address the environmental impact of rice cultivation.
CarbonFarm is using the ALU API and Gemini to automate field-level insights, including the delineation of individual fields, as part of programmes designed to support lower-carbon rice production. The company aims to support 2 million hectares of low-carbon rice by 2030.
The ability to accurately identify and monitor individual fields can also help improve access to agricultural finance.
Terrastack has used ALU and AMED APIs to develop a spatial intelligence platform that has mapped more than 140 million hectares of farmland. By reducing reliance on physical field visits, the platform is helping lenders and other stakeholders make faster and more accurate decisions across India’s agricultural ecosystem.
“Over 100 million farming households remain underserved because there is no reliable way to understand what is happening at the farm level. Google’s ALU and AMED models have enabled us to build a spatial intelligence platform that is helping transform fragmented land, crop, and income data into actionable intelligence for every farm in India,” said Aaryan Dangi, Co-founder and CEO, Terrastack.
“This creates the foundational infrastructure that enables lenders, insurers, governments, and agribusinesses to make better decisions, expand access to services, and strengthen the resilience of India’s agricultural economy,” he added.
Supporting India’s digital agriculture infrastructure
The models are also being integrated into public-sector digital agriculture initiatives.
In Telangana, the Agriculture Data Exchange (ADeX) platform is using ALU and AMED as part of efforts to support innovations serving more than 5 million farmers.
Among the applications being piloted is Krishivaas, which is designed to provide actionable, hyperlocal information on crop stress, crop-specific weather patterns and localised pest outbreaks.
“Historically, our planning relied on aggregate and largely descriptive datasets, making it challenging to understand water productivity at a local level. By integrating Google’s ALU and AMED models with KWRIS, localized weather and remote sensing data, field observations, and community participation, we have gained precision and timely crop intelligence,” said the Advanced Centre for Integrated Water Resources Management (ACIWRM), Water Resources Department, Government of Karnataka.
“This enables us to identify areas requiring productivity improvements and supports more informed decisions to improve water productivity and sustainable water resource management across our river basins,” it added.
Karnataka’s Water Resources Department is combining ALU and AMED with localised weather and remote-sensing data to support dynamic water management across the state’s 2.6 million hectares of irrigated land.
The approach is intended to provide more granular and timely information for managing water resources and improving water productivity.
Taking India’s agricultural AI infrastructure global
The applications are not limited to India. The United Nations Food and Agriculture Organization (FAO) is incorporating the models into its geoAI4stats initiative, which has received support from Google.org through the AI Collaborative: Food Security.
The initiative plans to integrate ALU and AMED into FAO’s global CROPGRIDS data repository, with the aim of strengthening agricultural monitoring and the availability of more granular data for agricultural sustainability.
“Better agricultural decisions start with better data. By combining AI, geospatial intelligence, and statistical systems, geoAI4stats will help countries access more timely and granular agricultural insights,” said Francesco Tubiello, FAO Senior Statistician and geoAI4stats Project Lead.
“With support from Google.org, we are bringing together advanced AI capabilities and FAO’s agricultural expertise to strengthen agricultural data as a global public good. This will help countries generate better insights for agricultural planning, sustainability efforts, and food security interventions while accelerating the transformation of agri-food systems,” he added.
The growing international use of the models illustrates how tools developed to address specific agricultural challenges in India can potentially become part of digital infrastructure supporting agricultural decision-making in other regions.
Measuring sustainable farming at field level
For CarbonFarm, the combination of satellite imagery and AI is also helping address one of the challenges associated with scaling climate-smart agriculture: measuring and verifying outcomes.
“For years, the biggest barrier to scaling sustainable rice cultivation wasn’t knowing how to reduce methane emissions. It was measuring, verifying, and paying for those reductions at scale,” said Aparna Raturi, Chief Operating Officer, CarbonFarm.
“By using Google’s ALU model to map individual field boundaries, enabling satellite-based verification of farming practices, and Gemini to give farmers real-time feedback as they implement water management techniques, we can now measure adoption, water outcomes, and methane emission reductions,” she said.
According to Raturi, the resulting data can help farmers adopt climate-resilient practices while supporting outcome-based incentive programmes built around measurable and verifiable results.
Building an ecosystem around agricultural data
The Telangana government sees shared agricultural data infrastructure as another potential area for AI-enabled innovation.
“One of the biggest challenges in agriculture is access to quality data. With ADeX, we set out to create a shared platform that Telangana’s government departments, universities, startups, and partners can build upon,” said the Information Technology, Electronics and Communications (ITEC) Department, Government of Telangana.
“By integrating Google’s ALU and AMED datasets, we are enabling capabilities such as field boundary delineation, crop stress analysis, early warning systems, and hyperlocal advisories. Together, these innovations are helping us create a stronger digital foundation for agricultural services and innovation across Telangana,” it added.
The model deployments point towards a broader shift in agricultural technology: from isolated applications to shared data infrastructure that can be used by governments, financial institutions, agribusinesses, researchers and technology companies.
From India-first models to global agricultural applications
Alok Talekar, Lead, Agriculture and Sustainability Research, Google DeepMind, who leads the AnthroKrishi team, said the growing use of the models reflects the potential of AI and geospatial data to address agricultural challenges across markets.
“Our AnthroKrishi team has been dedicated to supporting targeted agricultural solutions that both increase farm productivity and reduce climate impact. The growing application of our India-first AI models’ APIs to impact-focused solutions, ranging from farmer credit to crop advisory and policy decision-making, across both the Indian and global ecosystem encourages us in our approach,” said Talekar.
“As these models expand to support even more countries, we look forward to the immense potential they will unlock for key global priorities, from food security to agricultural resilience,” he added.
As adoption expands across sectors and geographies, the experience of ALU and AMED demonstrates how India-focused AI development can extend beyond its original context. By making model outputs available to the wider ecosystem, Google and Google DeepMind’s AnthroKrishi team are seeking to enable organisations to develop their own applications around agricultural data.
For an agrifood system facing simultaneous pressures around food security, climate resilience and resource efficiency, such shared digital infrastructure could play an increasingly important role in making agricultural decisions more timely, targeted and data-driven.
Also Read: Jefferies Maintains Buy Rating on UPL, Sees 14% EBITDA Growth in FY27
Global Agriculture is an independent international media platform covering agri-business, policy, technology, and sustainability. For editorial collaborations, thought leadership, and strategic communications, write to pr@global-agriculture.com
