Artificial Intelligence Will Predict a Good Harvest
A new AI system developed by researchers at Kuban State Agrarian University and partner agricultural companies is pushing Russia’s precision farming sector forward by predicting crop yields and helping farmers boost productivity
AI That Knows and Advises
Researchers and agricultural producers in southern Russia are completing trials of a neural network designed to enhance precision agriculture systems across the country. The technology predicts crop yields by analyzing fertilizer application schemes for each specific field.
The project is being developed by specialists at Kuban State Agrarian University (KubGAU) in partnership with Progress Agro Group and Prof Agro. The system is built on mathematical models that account for fertilizer types, application methods, soil characteristics, plant conditions, and growth dynamics.
More importantly, the neural network actively helps farmers improve future harvests by recommending optimized fertilizer strategies. Based on real data, the AI suggests application patterns tailored to individual zones of a field, often reducing fertilizer consumption without sacrificing yields. This is especially critical given the rising cost of fertilizers and fuel—optimizing inputs directly increases profitability and reduces soil stress.
Trials and Data-Driven Gains
To validate performance across different conditions and crops, the AI underwent extensive testing. Over two years, it was deployed on 17 fields, where it delivered yield increases of up to 6.3%. At the same time, fertilizer use declined—saving an average of 24 kilograms of fertilizer per hectare. Total fertilizer costs on trial fields dropped by approximately $20,600.
Unlike traditional precision farming tools, which rely mainly on productivity zones or vegetation indices, the new neural network integrates vast historical datasets and continuously updates its models. It accounts for chemical soil profiles, microclimates, and field-specific patterns that existing methods cannot fully capture.
According to Dr. Evgeny Truflyak, head of the Department of Operation and Technical Service at KubGAU, “Further development includes expanding the training datasets, adapting the technology for more crops and regions, creating full-cycle digital farm management systems, and developing software for large-scale deployment of AI tools in agriculture.”
New Crops, New Opportunities
The next stage of testing will assess performance on more demanding crops—sugar beet, soybeans, sunflower, and corn. These crops have completely different agronomic requirements, offering a rigorous evaluation of the neural network’s adaptive capabilities.
For Russia’s agricultural sector, the project represents a major step toward mature digital farming. Universities and industry partners are demonstrating the ability to build full-spectrum precision agriculture platforms, including variable-rate fertilizer systems, smart machinery for uniform seeding, autonomous navigation for planting and harvesting, and complete “Field History” analytics tools.
Once testing is complete, the KubGAU–Progress Agro–Prof Agro neural network will become a key component of Russia’s precision agriculture toolkit. With refinement, it holds export potential—particularly in regions seeking resource-efficient agricultural technologies, such as the CIS, the Middle East, Asia, and Africa.
