To create a more connected and intelligent global agricultural research ecosystem, CGIAR convened a Data Harmony Workshop at CIMMYT Headquarters in Tex Coco, Mexico from June 17-20, 2025. The main goal of the workshop was to develop a practical framework for ensuring data impartiality in all CGIAR centers. This effort is driven by an increased donor pressure on transparent data sharing and the need to prepare a vast dataset for the demands of artificial intelligence that uncovers new insights in agricultural science. By harmonizing the data, CGIAR aims to amplify its impact on research, generate efficiency and enhance partnership opportunities.
The hybrid event, which attracted participants from 12 CGIAR centres, is a DTA initiative, particularly under Work Area 1 (AOW1), focusing on building a robust data ecosystem for CGIAR.
Building a consensus across the center
Unlike past efforts where they struggled to gain traction, the four-day working session highlighted co-creation and consensus. This highlighted the narrative shift from strict “standardization” to more coordinated “harmony.” The focus lies in general guidelines rather than imposed standards, and emphasizes participatory processes that respect diverse data practices across the centre, based on previous efforts. This process included representatives of the alliance of Bioversity & Ciat, Cimmyt, Ifpri and IITA, as well as virtual contributions from Africanage, Cifor-Icraf, CIP, Icarda, Icrisat, Ilri, Irri, Iwmi and Worldhish.
Participants agreed that their top priority was to harmonize data disclosure rather than data collection. Two early domains (anomalous and socioeconomics and gender) were used to define “core variables,” to create harmony guidelines, and to test the process of designing pathways for adoption. These pathways include both mandate-driven approaches (backed by management and linked to funding requirements) and motivation-driven approaches (encouraging voluntary uptake by researchers through clear incentives).
Key results and output
Three main outputs were generated in the workshop.
- Draft Harmony Guidelines (v0.1) – A practical document to guide future data publishing, including naming conventions, metadata requirements, and attribution criteria.
- Core Variable Set – Baseline list of variables in agriculture and socioeconomic data. Designed to improve comparability and reusability.
- Two-Pathway Adoption Model – Combines top-down mandates with bottom-up motivations to promote compliance without burdening researchers.
The core variable is not the ceiling, but the starting point. “The center has the flexibility to add extended variables to meet specific research needs while aligning with the baseline for interoperability.
Why is it important?
While donor agencies are increasingly demanding data sharing and reusability as conditions for funding, AI-driven research tools require clean, standardized data to function effectively. Workshop participants reflected on why previous harmony efforts have become unstable. We measured the perception among scientists that lack of incentives, poor technical preparation, and that data sharing adds to workloads without providing clear benefits.
The purpose of this initiative is to create a shared sense of ownership and change it by embedding harmony in workflows, proposal processes, and monitoring frameworks such as CGIAR's PRMS. The ultimate vision is for all public CGIAR datasets to meet harmonious baselines, increase impact, reduce duplication, and enable cutting-edge AI applications.
Next Steps
Immediate priorities include finalizing the v0.1 harmony guidelines, completing core variables for additional domains such as crop breeding, livestock (including aquatic organisms), environment and climate, and developing communication and engagement strategies.
A call to action
As CGIAR sets foot in a new era of data-driven science, the workshop marks an important milestone. It can change not only how data is processed within the system, but also how agricultural research is conducted and how it is shared around the world.
