When a farmer discovers a better way to deal with drought, a village learns to manage its water more efficiently or a community finds a way to cope with climate shocks, that knowledge often remains where it was created. It may sit in a field report, an interview recording or simply in the experience of the people who live there.
Bengaluru-based Apurva.ai is trying to change that. Founded by Anand Rajan, the AI-enabled suite of products brings together community voices, institutional knowledge, and ecosystem intelligence to help the development sector, including funders, governments, NGOs, and ecosystem orchestrators, make sense of complex social challenges, make better decisions, and enable greater impact at scale. The venture is backed by Nandan and Rohini Nilekani.
Building Digital People Infrastructure
Rajan describes the idea as “Digital People Infrastructure” – public-good infrastructure where community voices serve as the north star for systems change. “Communities are proximate to the problem and hold the most context,” he says. Organisations accumulate programme knowledge, researchers generate evidence, and funders see across portfolios, but those perspectives rarely come together when decisions are being made.
The parallel Rajan draws is with UPI. Its significance was not merely that it made payments easier; it created common infrastructure on which banks, fintechs and businesses could build. Rajan believes the social sector could benefit from a similar layer that allows knowledge to move more freely. “Shared, public-good infrastructure can allow organisations to build on what already exists and focus on solving the problems they know best,” says Anand.
India provides a large test bed. Across more than half a million villages, organisations working in agriculture, livelihoods, health, and climate generate enormous amounts of knowledge. Yet much of it remains scattered inside project reports, field notes, interviews, and the memories of people who have spent years working with communities.
One of Apurva.ai’s projects is with the Foundation for Ecological Security (FES), helping create a trusted knowledge infrastructure by bringing together curated experiences from local communities working on shared challenges around land, water, biodiversity, and livelihoods. Community members from the villages that FES works with could ask questions in their native language and receive grounded, traceable responses drawn from the lived experiences of other community members rather than a generic AI response. This enables contextual, trusted peer to peer learning and demonstrates how AI can assist communities to learn from one another at scale.
Likewise in Tamil Nadu, the ministry of agriculture & farmers welfare, as a prototype, deployed Apurva.ai to document farmers’ innovations and lived experiences, make that knowledge accessible across languages, and enable farmers to learn from one another.
AI for grassroots impact
Apurva.ai is applying the same thinking elsewhere. With the SELCO Foundation, it is building a “core community brain,” connecting the experiences of practitioners, enterprises, financiers, technology providers, and users of energy solutions. The effort is to make the development sector’s diverse knowledge democratic, synchronised, and accessible to all.
In collaboration with SELCO Foundation and WASSAN, it has also developed Millet Mentor AI, an assisted AI platform designed to deliver experiential training and subject matter expertise to millet entrepreneurs, farmer producer organisations, self-help groups, and processors. It supports multilingual interactions and integrates systems thinking with design thinking for holistic guidance on millet production, processing, value addition, and business development.
With Dasra, meanwhile, Apurva.ai is working across documents, portfolio data, and conversations to help teams retrieve institutional knowledge and surface insights without manually sifting through large volumes of material. The use cases differ, but the problem is similar: decades of learning often remain locked inside organisations and programmes.
“Listen, learn, and act” is how Rajan describes the approach. First, listen to communities at scale; then connect what they are saying with institutional and ecosystem knowledge. Those insights can then help determine what needs to happen next.
The work is also moving beyond India, across Central and South America, Africa, and the Asia-Pacific. Its partnership with Rare, the international conservation organisation, offers one example. The platform captures and analyses community feedback, including recorded interviews, to understand how people are experiencing climate events and local interventions.
The geography may change, but the gap remains familiar: those closest to a problem often have the richest context, while those making decisions struggle to access it. “The greatest value is not automation,” Rajan says. “It is shortening the distance between what people are experiencing on the ground and the decisions being made about them.”
The ambition is still being tested. But if a farmer’s experience in one village can help another elsewhere, institutional learning can be used when it matters, and community feedback can change a programme before it is too late, AI begins to serve a different purpose, less a productivity tool and more shared infrastructure.