The Agrarian Reform and Agroecology Artificial Intelligence (IARAA) project aims to develop an artificial intelligence (AI) tool that can help expand and strengthen agroecology. Coordinated by the Landless Workers’ Movement (MST) and the World March of Women, and supported by the International Association for People’s Cooperation (Baobab), the project was conceived as an international initiative that can be developed together with grassroots organizations in other countries.
The first in-person workshop for the project brought together ten agroecology specialists from across Brazil along with the technical development team. It took place on November 3 and 4, 2025, at the Florestan Fernandes National School.
According to the MST, “through the IARAA project, grassroots movements are taking a stand, building an AI tool to expand democratic access to agroecological, popular, and traditional knowledge and wisdom. They are bringing a people’s perspective into a field that remains largely inaccessible: the development of digital technologies.”
Collective Development: Methodology and Principles
The way IARAA is being developed reflects the principles of collective organization that guide the movements behind it. While large AI companies often appropriate and profit from knowledge created by society as a whole, IARAA starts from the principle that knowledge is produced collectively and should be recognized as such.
The project is based on the understanding that agroecological knowledge has been developed over generations by peoples, communities, and grassroots organizations. Research institutions and universities have also played an important role in producing and organizing this knowledge. One of the project’s main challenges is bringing together this wide range of knowledge in written form so it can serve as the AI’s knowledge base. Today, that knowledge is scattered across many sources and, in some cases, exists only as oral tradition.
To establish the project’s technical and political foundations, the movements assembled a team of agroecology experts representing all regions of Brazil. This group helped build the knowledge base that powers the tool and developed the guidelines that shape its responses. These guidelines are intended to ensure scientific and technical rigor while also reflecting the productive, organizational, and political dimensions of agroecology. This ongoing collective process is essential to ensuring that the tool does not reproduce the logic of agribusiness or promote one-size-fits-all technological solutions. Instead, it is designed to support diverse agroecological practices that are rooted in local conditions and realities.
Building a tool around these principles requires new approaches to technology development. On one hand, it involves training and technical education for grassroots movement activists so they can actively participate in the design, development, and evaluation of the system. On the other hand, it requires programmers and developers to deepen their understanding of the political, theoretical, and practical foundations of agroecology.
This helps ensure that those principles are reflected in the tool’s features, architecture, and user experience, strengthening the capacity of agroecological movements to organize and collaborate. The project therefore treats technical development and political vision as inseparable. Rather than separating technology from its social purpose, IARAA seeks to embody agroecological principles in its design while continuing to evolve through use and collective participation.
The Architecture
IARAA uses a technological architecture based on Retrieval Augmented Generation (RAG), which combines information retrieval with natural language generation. When a user submits a question, language models first interpret the request and identify its key concepts and context. The system then searches specialized knowledge bases that have been built and validated by grassroots movements to find relevant information. Based on this material, it generates a response that combines technical accuracy with clear and accessible language.
Unlike commercial chatbots that often simplify answers and promote standardized approaches, IARAA is being developed to account for differences in biomes, farming systems, forms of social organization, and local material conditions. The goal is not to replace existing technical or popular knowledge, but to strengthen it and make it easier to share across territories and generations.
