Innovation is not the same thing as digitalization: grassroots innovation, appropriate technology, and the knowledge infrastructure agroecology needs.

As the world's institutions meet to discuss "smart farming", data and artificial intelligence, one question is rarely on the agenda: whose intelligence, and whose farming? The dominant story of technology in agriculture is by now familiar — sensors and satellites, private platforms and predictive models, expensive tools that promise precision and scale. That story has real achievements in particular areas, such as efficiency and pest management. But as speakers at CFS High-Level Forum on Artificial Intelligence, Digitalization and Data Governance (Rome, 30 June 2026) pointed out, there is still no clear evidence that digitalization and AI improve food security and nutrition overall, and many of the risks have not been assessed.
It is also not the only story. For the vast majority of small-scale food producers, it is not their story at all. It is not the story of the rural workers, smallholders, fishers and pastoralists who grow food every day, manage risk as they go, and develop solutions that come out of their own knowledge, culture and living territories. Their perspectives are rooted in a place. Their problems are specific. And their ability to solve those problems themselves is too often ignored and pushed to the margins.
Agriculture has always been technological. The plough repaired and improved in a village workshop. The seed selected and exchanged across generations. The timing of an irrigation channel, held in a farmer's memory. The intercrop that stops a pest without a single molecule of pesticide. These are technologies. They are simply a different kind: low-cost, repairable, adapted to the place they are used, and owned and governed by the people who use them. Nearly forty years of work by the Honey Bee Network in India starts from one uncomfortable figure: about 95 percent of grassroots innovation is never written down, never shared and never built on. The intelligence is already in the field. What is missing is the infrastructure that would let it travel on the innovators' own terms.

Different innovation models, one quiet question
This matters more now than ever. Ten years ago, when agroecology entered FAO's halls through symposiums, civil society organisations warned that powerful interests might reduce it to one more tool in the industrial toolbox. The same risk applies to today's enthusiasm for innovation — a word that, in practice, has come to mean digital. When "technology for agriculture" is defined narrowly, as high-tech, data-hungry and privately owned, agroecology is described as the low-tech past. And a quieter question disappears from the room: who owns agricultural knowledge, and who benefits from it?
That question decides everything else. Data governance decides who collects, controls, shares and profits from the information that every digital system runs on. A technology can be impressive and still take knowledge away from the people who produced it. It can be simple and still leave them in control. The measure that matters is not how advanced a tool looks. It is who holds the keys.
This is not a contest between two camps. Most food producers already work with a mixture: a phone, a paper notebook, a neighbour's advice, a satellite image, a variety kept for thirty years, a spreadsheet at the end of the season. There is a great deal in between, and that is where most farming actually happens. The question is not which side of a line a tool sits on. It is on whose terms it is used, and who decides.
Agroecology already has a principle for this. Among FAO's 10 Elements of Agroecology is one that the current excitement tends to overlook: the co-creation and sharing of knowledge. Agroecological knowledge is not delivered from a laboratory to a passive receiver. It is produced together, it depends on context, it holds many valid answers at once, and it keeps changing. This is exactly the kind of knowledge that industrial data systems handle worst. They are built to reduce variation to uniform categories, to optimise for scale, and to move information upwards into fewer and fewer hands. Agroecology needs the opposite: technology that can hold many different answers without making them the same, and that lets knowledge travel without taking it away from its source.

The assembly, not the archive
This is the work of the Grassroots Innovations Assembly for Agroecology. GIAA is not an institution that collects the world's farming knowledge into one central store. It is a federation — a coming together of grassroots innovation networks: the Honey Bee Network's decades of documented farmer ingenuity, Prolinnova's participatory research, farmer-engineers building open-source machines in their own workshops, and food sovereignty movements across Africa, Asia, Europe and the Americas. Its founding principle is easy to say and hard to do: federation without flattening. Networks connect and share through a commons, each one keeps control of what belongs to it, and none is dissolved into a single platform that speaks for all.
The ethical rules come, fittingly, from the grassroots networks themselves. The Honey Bee Network puts them in three words: recognition, respect, reciprocity. Recognition means knowledge is never used without naming its source, and that source can be a community as easily as a person. Respect means consent that can be given, reviewed and withdrawn, recorded in whatever form the knowledge holder can use — spoken, written or witnessed — not a signature collected once and relied on forever afterwards. Reciprocity means you do not take knowledge unless you are ready to give: the network sends every contributor a collection of what others have shared. In a time of data extraction, these are not old-fashioned good manners. They are a governance model more advanced than most of what the technology industry has produced.

A machine that runs on sunlight
What does this model look like as an actual machine? Consider the Farm Hack Box: a small, low-power computer — it can run on solar power, using about as much electricity as a light bulb — that holds a farm's records, maps and knowledge, works without an internet connection, and is owned outright by the farm. On its own it is a modest device. Connected with others, it becomes something larger: a node in a knowledge commons that no company hosts and no single actor controls.
The organising idea is older than any computer: the library. Libraries have shared knowledge across institutions for centuries without ever selling each other a licence to their catalogues. The catalogue — the description of what a collection holds — is shared freely, because you cannot use what you cannot find. The terms of use apply to a particular work, at the moment it is exchanged. Grassroots innovation infrastructure can work the same way, in layers. A community shares a catalogue of what it holds: a description of a practice, a seed variety, a tool, a map. It does not hand over the knowledge itself. More detail is shared by agreement. The knowledge moves only under the community's own conditions. This is what makes it safe to take part: you can say what exists in your world without being made to give it away. Older encyclopedias once gathered the knowledge of craftspeople and gardeners into a shared inheritance; this time, the contributors keep authorship of their own entries. And this is what could finally make the missing 95 percent visible — not by extraction into someone else's database, but by federation into a commons the innovators govern.

Appropriate technology, brought up to date
Fifty years ago the appropriate technology movement — Schumacher's "small is beautiful", Illich's insistence on tools that widen rather than narrow human autonomy — argued that the right technology is one a community can understand, repair and control. Grassroots innovation carries that idea into the digital age, where "appropriate" must now include appropriate data governance. The networks are building their own instruments for this: ethical technology assessments that let a community weigh up a tool before adopting it. Does it leave control of our data with us? Can we see how it works? Can we leave, and take our knowledge with us? Communities can then refuse the tools that fail. This is not being against technology. It is communities doing for themselves what the market will not do for them: deciding, on their own terms, which technologies deserve a place in an agroecological future.
It also points to a different idea of scale. The investment logic behind "smart farming" scales the way industry always has: vertically, by concentrating tools, data and decisions in fewer and larger hands. Agroecological knowledge scales differently: horizontally, farmer to farmer and network to network, becoming richer rather than more uniform as it spreads. A federated commons is built for the second kind of scaling. It does not ask a million farmers to send their knowledge to one place. It lets a million places find and learn from each other while staying themselves.

Inventing the future, on our own terms
The choice is not between a high-tech future and a nostalgic past. It is a choice about ownership and governance: whether the knowledge infrastructure of our food systems is built to concentrate power or to spread it. The most hopeful proposition of the Grassroots Innovations Assembly for Agroecology is also its most challenging: that farmers, fishers, pastoralists and Indigenous Peoples can build, own and govern the technology and the digital commons of their own food systems — and that doing so is not a retreat from technology, but the most mature use of it.
Which brings us back to the words. It is worth asking what we mean by "smart", and what we mean by "intelligence". If those words describe a system that observes, learns, adapts to local conditions and improves over time, then a farming community that watches, tests, records and passes on what it learns across generations is the older and more demanding version of both. That kind of intelligence runs on sunlight and on consent. It is built in workshops and villages as much as in laboratories. And it is already being built by the people with the most knowledge and the least recognition. As this column has said before, the best way to predict the future is to invent it. Agroecology's inventors are already at work. What they need now is a commons of their own — one they build, own and govern.

The Grassroots Innovations Assembly for Agroecology (GIAA) is a federation of grassroots innovation and food sovereignty networks working to build open, appropriate and farmer-governed knowledge infrastructure, technologies and innovations for agroecological transition. This piece draws on the shared work of its member networks, including the Honey Bee Network, Schola Campesina and Farm Hack.

A Germinate! guest column by Dorn Cox (Farm Hack US), with contributions from Caroline Ledant (Schola Campesina), members of the Grassroots Innovations Assembly for Agroecology (GIAA).