AI for farmers: UAE develops free tool trained to solve crop-specific issues
The UAE has developed a specialised artificial intelligence tool designed to support farmers by answering crop-specific agricultural questions with greater accuracy than general-purpose AI systems. The platform, known as AgriLLM, is fully open-source and free to use, allowing anyone to access, modify, or build upon it.
Developed by Abu Dhabi-based ai71 in collaboration with 15 international organisations — including CGIAR and the Gates Foundation — the tool aims to address a major global challenge: limited access to reliable agricultural guidance. According to the UN’s Food and Agriculture Organization, nearly 75 per cent of family farmers worldwide lack consistent technical support.
AgriLLM is a large language model specifically fine-tuned for agriculture, using curated datasets rather than broad, multi-domain information. Mehdi Ghissassi, Chief Product and Technology Officer at ai71, said the model was designed to prioritise factual accuracy and practical usefulness over lengthy responses.
Unlike general AI platforms such as ChatGPT, which are trained on diverse internet data, AgriLLM relies on verified agricultural sources. Its training dataset includes more than 350,000 agricultural documents, 50,000 research papers, and 120,000 real-world farming questions with validated answers.
Internal evaluations conducted by ai71 indicate that AgriLLM delivers factually correct answers about 30 per cent more often than GPT-4o when responding to agricultural queries. The model is designed to provide concise, evidence-based guidance, reducing the risk of misinformation that could negatively impact farming outcomes.
“In agriculture, an incorrect answer can have serious consequences,” Ghissassi said. “That’s why accuracy and traceability of information were central to how the model was developed.”
The system is capable of addressing crop-specific issues, regional growing conditions, and climate-related challenges. Rather than offering generic advice, AgriLLM can reference relevant research and adapt its responses based on factors such as soil type, climate, and geographic location.
The platform also supports progressive refinement: broad questions generate general guidance, while follow-up queries result in increasingly precise recommendations. This makes the tool useful not only for farmers, but also for agricultural extension officers, researchers, and policymakers.
By making AgriLLM freely available, developers hope to strengthen agricultural resilience, improve productivity, and support sustainable farming practices globally — particularly in regions where access to expert advice remains limited.
