A service providing agricultural technology information to farmers. The Agricultural Encyclopedia provides a chatbot-based service trained (RAG) on agricultural technical manuals and related data held by the institution. Agricultural Education delivers customised educational video content tailored to users based on membership information. Farm Planning provides necessary information (e.g. production, sales, cultivation) according to different farming stages (preparation, entry, growth stages).
Name in original language
AI이삭이
Initiative overview
Selected as a “Public Specialised Sector” project under the Ministry of Science and ICT’s 2024 Hyper-scale AI Programme, the project developed an agricultural AI agent between July and December 2024. In partnership with the private sector, the initiative built a HyperCLOVA X-based system that draws on agricultural knowledge and is continuously updated.
The consortium comprises:
- the Rural Development Administration, responsible for content;
- Naver Cloud, responsible for the LLM and platform; and
- Didim365, acting as the AI specialist.
A user pilot conducted from November to December 2024 informed improvements. Pilot operation and the official launch followed in September 2025. Response accuracy improved through additional data collection and expert feedback, reducing the need for multiple searches by delivering immediate, evidence-backed answers.
Key activities
- Agricultural Encyclopedia: Trained on 6,307 technical manuals and approximately 50,000 datasets, including variety information, weekly updates and field cases. The service minimises hallucinations, and voice question-and-answer functionality was added in November 2025. The service is also available through the Nongsaro portal.
- Agricultural Education: Recommends region- and crop-specific instructional videos and is trained on 1,768 technical videos and scenarios.
- Farm Planning: Offers crop recommendations, expected income and profitability strategies, covering ten sectors and 117 items, supported by ten years of income data across 54 categories.
- Real-time consultation: Available through Nongsaro, with 5,881 cases recorded from March to August 2024 and 5,264 cases from March to August 2025.
- Support for extension-centre operations: Delivered through ASTIS, using extension reports, guidelines and master plans.
- Safety integration: Connects with Farmers Safety 365 to support accident prevention and training.
- AI management consulting: Combines the Agricultural Encyclopedia and Farm Planning functions, with documented improvements at the farm level.
- AI education assistant: Under development to recommend courses through e-HRD.
Other relevant details
Results, outcomes and impacts: A survey of 146 young farmers found that approximately 90% were satisfied with the service. Respondents requested more substantive on-site decision support through richer and more comprehensive responses and a wider range of services. The 2026 enhancement plan includes redesigning the mobile application, using device functions such as voice, cameras and sensors, developing individualised RAG systems, adding a voice assistant capable of recognising regional dialects, and simplifying registration and login to improve usability for older farmers.
Challenges and lessons learned: Agricultural production cycles extend over long periods. Poor initial decisions or delays in applying appropriate interventions can cause severe crop losses affecting both farmers and consumers. Information must therefore be delivered promptly and accurately to support farmers’ decision-making and reflect rapidly changing conditions in the field. “AI Isak” was trained on diverse, authoritative agricultural data held by the Rural Development Administration to reduce the risk of inaccurate information reaching farmers. A data-learning system was also established to incorporate the latest information into the system promptly. Through the 2026 enhancement project, the initiative is improving service UI and UX and adding a voice assistant to make the service easier to use for an increasingly ageing rural farming population. Conditions for success include a dedicated unit with expertise in both agricultural realities and advanced AI, multidisciplinary expert participation, and sufficient capacity and budget to implement field-driven features through pilot services.
Future plans: By 2026, the initiative plans to enhance usability and functionality by redesigning the mobile application to strengthen its AI features, using device capabilities such as voice, cameras and sensors, and developing individualised retrieval-augmented generation systems for personalised support. It will also add a voice assistant that recognises regional dialects, simplify registration and login for older farmers, introduce AI-based image recognition for pests and diseases, and integrate consumption and distribution data into Farm Planning to support management strategies that account for weather-related disasters and market changes. By 2027, the platform is expected to expand its training domains using Rural Development Administration data, develop AI systems for preventing crop disasters related to weather, pests and soil, and establish a real-time AI agricultural-disaster monitoring system.
