FROM PILOTS TO PROOF: WHY THE FUTURE OF AI IN INDIAN AGRICULTURE WILL BE MEASURED IN FARMER DECISIONS, NOT APPS
A panel discussion on "Digital Agriculture and Food Systems Transformation: AI, Evidence and Partnerships for Sustainability, Resilience and Inclusion" convened by the International Food Policy Research Institute (IFPRI), the International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), and the CGIAR Science Program on Policy Innovations in Bangalore on 17 June 2026.
India is undergoing a significant transformation in “how it farms”. The country is moving from tradition-led cultivation towards a data-driven, precision-based ecosystem, one built on satellites, sensors, drones, weather stations and farm machinery feeding decisions at every stage of the agricultural value chain. Underpinning this is a deliberate act of public engineering: a digital foundation anchored by the Digital Agriculture Mission and AgriStack, designed to deliver verified, targeted services to millions of farmers. India now ranks third in the world in overall AI competitiveness, according to Stanford University's 2025 Global AI Vibrancy Ranking. However, competitiveness alone is not a measure of the outcomes, and this distinction sat at the heart of this panel discussion.
The more capable technology becomes, the sharper the real question grows: capability for whom, and proven how? It was to examine exactly this that IFPRI and ICRISAT, working through the CGIAR Science Program on Policy Innovations, brought together agri-tech entrepreneurs and developers, scientists, extension leaders and leading policy think tanks in Bangalore.
The debate has moved on
Setting the frame, panel chair Prof. Ramesh Chand, Distinguished Professor, Indian Council for Research on International Economic Relations (ICRIER) and Former Member, Niti Aayog, made it clear that the relevance of AI in agriculture, is no longer up for debate - the real question is whether India is moving fast enough to adopt it, and whether it can move from isolated pilots to mainstream adoption. He issued a caution that would echo through the rest of the session: as AI transforms agriculture, it must narrow rather than widen the divide between technologically advanced farmers and ordinary smallholders.
Dr. Mamata Pradhan, Senior Research Coordinator at IFPRI, opened by situating the discussion within a remarkable structural shift. India's agri-food tech sector has emerged as one of the world's largest agricultural innovation ecosystems, expanding from fewer than 100 startups in 2000 to 24,534 firms by 2025. This growth mirrors a global surge in agritech entrepreneurship and venture capital investment, and it is no coincidence that understanding digital agriculture in India today is inseparable from understanding agritech startups, which have become important mechanisms for agricultural modernization, innovation diffusion and productivity growth.
A cardinal mistake in digital agriculture, according to Dr. Pradhan, is the failure to locate digitization where the value actually is. Too many initiatives digitize peripheral processes while leaving the core of the agricultural value chain untouched. Digital agriculture is most effective when it is anchored to markets and integrated with extension systems, not when it sits beside them. Many digital agriculture initiatives struggle to move beyond pilot projects precisely because they are scaled before being rigorously evaluated. Getting digital agriculture into the value chain – where value is created, growth happens, and the future lies – must be the organizing principle.
In India, IFPRI served as technical advisor to Farm Vaidya, an AI startup developing a generative AI voice agent for Telugu-speaking smallholder farmers in south India. Building the underlying speech technology involved months of fieldwork, with more than 30,000 hours of conversational farm audio collected across India's 10 major languages and over 5,000 previously undocumented vernacular agricultural terms documented. The experience underscored Dr. Pradhan’s point that smart farming ultimately needs smarter systems grounded in farmers’ realities.
The promise of digital agriculture, according to Dr. Shalander Kumar, Deputy Global Research Program Director of the Enabling Systems Transformation Program and Cluster Leader, ICRISAT, lies not in building more platforms but in turning data into timely, actionable insight for farmers, for extension systems, and for policymakers. He set out three pillars for making that promise real, and they form the spine of how we read this entire transformation:
Evidence – measuring real-world outcomes, not merely technology adoption.
Climate intelligence – enabling proactive, context-specific decisions rather than reactive ones.
Inclusive scale – ensuring innovations reach smallholders, women farmers and underserved communities through trusted institutions and localized delivery.
IFPRI's own work in this space is already putting these pillars into practice. Through the Generative AI for Agriculture (GAIA) initiative, led by IFPRI in partnership with Centre for Agriculture and Biosciences International (CABI), Digital Green, the University of Florida, and SCiO, the institute is working to enhance the efficacy, reliability, and contextual relevance of AI-generated agricultural advisories for small-scale producers in the global South. Building on years of joint research on digital extension, IFPRI and Digital Green are deepening their collaboration to test AI innovations for smallholder farmers by focusing on user testing of Digital Green's FarmerChat application, an AI-powered assistant that provides farmers with free, localized, and climate-smart agricultural advice in their own languages, using text, video, voice, and images. Phase II of GAIA (2025-2027) refines these tools with a focus on usability, trust, and inclusion ensuring that AI reflects how farmers and extension agents interact in real-world settings.
Further, through the CGIAR India Policy Innovation Hub, IFPRI and ICRISAT are pooling their strengths across evidence generation, climate-smart agriculture, digital advisory and policy engagement to pursue this vision. The thread running through all three pillars is the same: technology is the beginning of the story, not the proof of it.
First the rails, then the intelligence
If evidence is the destination, data is the track. Mr. Jagdish Babu, Chief Operating Officer, Ekstep, offered the session's most durable image, describing digital public infrastructure as "data on rails." AI, he argued, can only ever be as good as the digital ecosystem beneath it. Standardized farmer registries, land records, crop-sown information, weather data and interoperable soil datasets are the foundations on which any credible AI application must be built.
His example was Maharashtra's MahaVistaar platform, which resisted the temptation to lead with sophisticated predictive models. It concentrated instead first on building a robust data grid and on making existing agricultural knowledge genuinely accessible to farmers. Only with that infrastructure in place does the next phase - personalized advisories tailored to a farmer's own land, crops, soil and local weather - become possible. The sequencing is the lesson: rails first, intelligence second.
AI as co-pilot, not replacement
For all its potential, the panel was firm that AI is not a substitute for agricultural expertise. These systems are probabilistic by nature and demand continuous validation by people who know the field.
Mr. Siddhant Panda, NLP & Generative AI Specialist at Soket AI Labs, framed AI's real value as integration – its ability to weave together weather, soil health, satellite imagery, crop conditions and local knowledge into recommendations a farmer can act on. AI, he suggested, should serve as the "reasoning layer" connecting India's growing digital agriculture ecosystem, but only if its models are trained on Indian agricultural conditions rather than borrowed from elsewhere. Just as important, human expertise must be preserved: rather than replacing extension workers and specialists, AI should function as a "co-pilot," helping experts validate recommendations and build farmer trust.
Dr. Dinesh Balam, Head - Food Systems, AI and Regenerative Agriculture, Bharti Institute of Public Policy, Indian School of Business (ISB), pushed the conversation onto harder economic ground, asking whether advisory services alone can sustain viable business models. AI adoption, he argued, must be driven by clear value creation for farmers – not by novelty. He pointed out the importance of going beyond advisory platforms altogether, towards food waste valorization, circular economy approaches and ingredient science, urging that future investments address broader food-systems challenges rather than digital advisory alone.
Inclusion is the test, not the footnote
Another crucial aspect of the discussion was that digital tools, on their own, do not fix old problems. Dr. C. P. Gracy, Former Senior Professor & Head (Agricultural Economics), University of Agricultural Sciences, underscored the enduring role of Farmer Producer Organizations (FPOs), noting that AI can sharpen access to market intelligence and weather information, but farmers often remain constrained by limited storage, credit, transportation and bargaining power. Translating innovation into tangible gain, she argued, requires FPOs, bundled service delivery, multilingual advisories and supportive regulation working together.
Dr. Kishore Kumar, Associate Scientist for Digital Innovations at ICRISAT, drew attention to another important aspect. The real test of digital advisory, he said, is not whether a farmer receives it – but whether it helps that farmer make a timely, useful decision. Advisories become actionable only when tied to crop stage, local weather and the farmer's immediate decision window.
His point about delivery connects directly into the question of building farmer capability over time: digital tools can reach a phone, but trust is built locally. Last-mile delivery through extension workers, Krishi Vigyan Kendras (KVKs), FPOs and farmer networks is therefore not an add-on, but a precondition and each such interaction is where a farmer's confidence and capacity to act on data is steadily strengthened. Advisory systems must also reckon with practical constraints: input availability, water access, labor and market conditions. On the question of scaling, Dr. Kumar was unambiguous – success cannot be measured by outreach numbers alone, but by evidence of decision change, climate-risk reduction, and the genuine inclusion of women farmers, smallholders, tenant farmers and the digitally excluded. Real scale, he concluded, means better decisions, lower risk and stronger resilience.
From innovation to impact
The session found its fitting close in the words of Panel Chair, Dr. P.K. Joshi, President, Agricultural Economics Research Association (AERA) and Vice President, Social Sciences, National Academy of Agricultural Sciences (NAAS), who said that real progress depends on generating rigorous evidence of impact, embedding climate intelligence into advisory systems, insisting on inclusion, and building durable partnerships between governments, research institutions, technology providers, universities and farming communities.
The panelists agreed that the true measure of digital agriculture lies not in the number of applications built, but in the outcomes it delivers for farmers.
AI, in the end, is not a standalone solution. It is a powerful tool – and only when it is anchored by robust institutions, evidence-based policymaking, inclusive governance and the patient, continuous work of building farmer capability does it help construct food systems that are genuinely more resilient, more sustainable and more farmer-centric. The technology has arrived. The task now is to ensure that it delivers meaningful impact for our farmers.
This blog was authored by Ishita Sirsikar, Communications Associate at IFPRI, with contributions from Anisha Mohan, Communications Specialist at IFPRI, and Dr. Mamata Pradhan, Senior Research Coordinator, Development Strategies and Governance (DSG) Unit, IFPRI. The blog is based on the discussions and perspectives shared by the panelists during the panel discussion, "Digital Agriculture and Food Systems Transformation: AI, Evidence and Partnerships for Sustainability, Resilience and Inclusion" in Bangalore on 17 June 2026.
