Around 60% of Indian farmers continue to grapple with fragmented landholdings, erratic weather, and limited access to data-driven insights. With over 70% of rural households dependent on agriculture, productivity remains heavily constrained. As climate challenges intensify, Google DeepMind is stepping in to bring greater precision, scale, and intelligence to the country’s farming ecosystem.
The company has announced that the APIs of its foundational AI models for agriculture, first launched in India, are being expanded to Asia-Pacific countries, including Malaysia, Indonesia, Vietnam, and Japan.
The Agricultural Landscape Understanding (ALU) API, made available to Indian developers in October 2024, and the Agricultural Monitoring and Event Detection (AMED) API, released in July 2025, are now open to trusted testers across the region.
In an exclusive discussion with AIM, Alok Talekar, lead of agriculture and sustainability research at Google DeepMind, and Avneet Singh, product manager for Google partner innovation, shared insights into the company’s ambitious plans.
Talekar drew a comparison between India’s Aadhaar system and DeepMind’s AI agriculture models. “Aadhaar was built in India, but it has now found its way to many other countries,” he said. He added that the challenges India faces in agriculture are not unique.
“Most large-scale solutions usually target big farm owners in places like the US and don’t address the needs of smallholders in the Global South. Our motivation is to serve all of these farmers and help them benefit, wherever they are.”
Understanding ALU and AMED
The APIs use remote sensing and machine learning to provide insights that can support targeted agricultural solutions. Talekar said that DeepMind’s work focuses on applying AI to help farmers and agricultural stakeholders make data-driven decisions. “Our goal has been to support data-driven decision-making for these ecosystem players, like governments distributing subsidies, banks giving loans, and companies providing agricultural inputs,” he said.
The ALU API identifies fields, water bodies, and vegetation boundaries, while the AMED API delivers field-level insights on crop types, sowing and harvest timelines, and agricultural events, updating data approximately every 15 days. Together, they provide a base for precision agriculture tools, resource optimisation, and improved farm management.
“Timely access to information is critical. You can’t make decisions about fertiliser distribution or irrigation based on outdated data,” Talekar said.
To put this into practice, Singh said that farmers in India access DeepMind’s AI agriculture tools through a network of partners rather than directly. For example, the Government of Telangana has built a platform called Adex, which connects data users and providers on an open network to build agricultural datasets for farmers.
TerraStack, incubated at IIT Bombay, uses the APIs for the digitisation of land records.
In India, the APIs are also being integrated into platforms such as Krishi DSS, developed by Amnex for the Department of Agriculture and Farmer Welfare, and are being used by organisations including the Council on Energy, Environment and Water (CEEW), Vassar Labs, and Sugee.io.
Applications range from policy planning and crop advisory to income support mechanisms, climate-smart agriculture, and rural credit management.
Amnex Infotechnologies is developing an integrated Krishi Decision Support System (DSS) platform using Google’s ALU and AMED APIs to deliver real-time, data-driven insights for agricultural planning.
Mihir Dakwala, business head for Agritech and data fabric at Amnex, told AIM that the integration has shown promising results in early tests, thanks to better alignment between the analytical logic units and environmental data modules.
“This has improved both accuracy and stability in initial runs, showing strong potential for scaling,” he added.
Technology and Future Plans
When asked about the technology behind the models, Talekar highlighted Google’s decades-long investment in geospatial understanding and high-resolution satellite imagery. “Our models inherently depend on remotely sensed information, either from licensed or public satellite data,” he said. He shared that the APIs are served free of cost to trusted partners, who validate the outputs through localised ground checks.
Regarding the possibility of integrating the models with Gemini, Singh said, “There aren’t any decided plans on the roadmap for integration with Gemini and our other Gen AI suites. However, if there are downstream applications that choose to combine Gemini with the output, that is obviously more than welcome.”
He further added that Google encourages partners to combine its various models and services to develop their own applications, describing it as a strong use case. However, DeepMind’s focus remains on delivering model outputs through APIs.
When asked about Google DeepMind India’s global role, Talekar highlighted inclusivity. “Our focus is on inclusive AI, addressing language diversity, healthcare, sustainability, and energy efficiency for AI models. The work in India contributes to DeepMind’s broader mission to build AI that serves global challenges,” he concluded.
The post Why Asian Farmers are Turning to Google DeepMind appeared first on Analytics India Magazine.
