Google focuses on India's farms and languages in new AI initiatives

Google unveils open-source AI initiatives targeting India's agriculture and linguistic diversity, aiming to enhance farming productivity and cultural representation in AI models, boosting partnerships

Google launched open-source artificial intelligence (AI) initiatives on Thursday, targeting India’s agriculture sector and cultural representation in AI models. The tech giant introduced its Agricultural Monitoring and Event Detection (AMED) Application Programming Interface (API), which tracks crop and field data across India to help developers create farming productivity tools. Researchers at Google DeepMind also partnered with IIT-Kharagpur through the company’s Amplify Initiative to build datasets capturing India’s linguistic and cultural diversity for integration into large language models (LLMs).

 

These developments build on Google's sustained investments and commitment to AI research that assist real-world impact across critical areas while also supporting India’s AI-focused ambition.

 

"We've been inspired by the solutions India's innovators have unlocked with these capabilities, demonstrating AI to be a powerful catalyst for multiplier impact and unprecedented effectiveness,” said Dr Manish Gupta, senior director for India and Asia Pacific (Apac) at Google DeepMind, during a roundtable in Bengaluru.

 

Google DeepMind and the Partnerships Innovation team have developed AMED API to improve agricultural monitoring across India. Building on the company’s Agricultural Landscape Understanding (ALU) API, the new tool uses machine learning, crop labels, and satellite imagery to identify crop types, field sizes, and sowing and harvesting dates. It also offers three years of historical data to track agricultural activity at the field level.

 

These insights aim to help develop AI-driven solutions that improve farm management. This is done by addressing crop-specific needs such as soil, water, growth patterns, and climate while also forecasting harvest volumes.

 

Agriculture and Sustainability Research Lead of Google DeepMind Alok Talekar said the firm is working on accelerating crucial shifts, transforming broad insights to granular, real-time data. “So that increasingly impactful solutions not only translate into benefits for India's farmers, but also bolster the nation against rising climate risks,” said Talekar.

 

TerraStack, a startup incubated at IIT-Bombay, has used ALU API to build a rural land intelligence system. The aim is to support rural lending, land record modernisation, and determine the vulnerability of farms to climate risk. It is exploring AMED API for a rural-lending use case.

 

“These APIs are helping standardise and transform previously unorganised and unusable data into solutions for one of India’s most critical sectors,” said Aaryan Dangi, co-founder and chief executive officer (CEO), TerraStack.

 

Linguistic diversity

 

Google’s Amplify Initiative seeks to improve LLMs by incorporating localised data — including regional languages, dialects, and cultural nuances — missing from current AI training. Partnering with IIT-Kharagpur, the project will develop high-quality, hyperlocal datasets capturing India’s linguistic diversity.

 

The open-source datasets aim to help developers create AI tools that better serve Indian language users. Data collection follows a community-driven, expert-vetted process to ensure responsible handling and reduce bias.

 

After a pilot in Sub-Saharan Africa, producing 8,000 annotated queries across seven languages, the India phase will focus on healthcare and safety topics in multiple Indic languages.

 

 

“We are meticulously building the rich, hyperlocal context and cultural understanding that transforms raw information into profound knowledge,” said Madhurima Maji, lead program manager for India at Google Amplify Initiative.

 

Dr Mainack Mandal, assistant professor, IIT-Kharagpur, said the collaboration opens a new chapter in global AI development.

 

The Amplify Initiative builds on Google’s broader push to improve Indian language and cultural representation in AI, alongside its flagship Project Vaani. Developed with the Indian Institute of Science (IISc) Bangalore, Project Vaani has released its second-phase Indic speech data through Bhashini and Hugging Face.

 

So far, the initiative has contributed nearly 21,500 hours of speech audio, and 835 hours of transcribed data across 86 languages, collected from over 112,000 speakers in 120 districts. The open-source data aims to support AI tools tailored for India’s linguistic diversity.

 

“This support fuels our continued investments in language and culture research, and drives us to make our foundational models, on which India is building its AI ambition, more effective and efficient in processing Indian languages,” said Dr Partha Talukdar, language research lead, Google DeepMind.

 

Google said its AI models are being used across sectors in India, from improving maternal health programmes and streamlining patient-care to supporting agri-tech solutions and advancing the country’s sovereign AI efforts.

 

The AlphaFold Protein Structure Database — developed by Google DeepMind and now used by over 150,000 researchers in India — is aiding work on complex diseases such as cancer and autoimmune disorders.

 

Google said, with a focus on collaboration and ecosystem-driven innovation, it aims to drive broad, real-world impact through AI in India.