[Interview] From farm machinery app to farming AI… Daedong Connect’s “AI Daedongi” receives 170,000 questions

A farm machinery management app used to check where your tractor is and when consumables need replacing has turned into a farming consultation app. “Daedong Connect,” created by Daedong for customers who purchased its farm machinery, introduced “AI Daedongi,” a generative artificial intelligence (AI) consultation service for agriculture, in April 2024. Since then, it has expanded into a farming app equipped with an AI farm work journal, early warnings for weather disasters, regional subsidy information, and agricultural product market prices.

The user composition has also changed. Subscribers, which numbered 30,000 in July 2025, surpassed 60,000 in September 2026, just one year and two months later. Of subscribers, 89% are farmers who do not own Daedong farm machinery. By age group, those aged 55 and older account for 51%, more than half, followed by ages 45–54 at 23%, ages 34 and under at 14%, and ages 35–44 at 12%.

We spoke with Lee So-hyun, assistant manager on Daedong’s Connected Business Team, who is in charge of Daedong Connect’s development, service operation, data analysis, and more, about the background to the service expansion and future tasks.

A farm machinery app opened only during busy farming seasons becomes a year-round app by adding farming information

Daedong Connect was originally a service for customers who own Daedong farm machinery. Its core functions were farm machinery management features such as checking the location of one’s farm machinery, remote fault diagnosis, consumables management, and remote starting. Assistant Manager Lee said, “With farm machinery functions alone, users only open the app during busy farming seasons in the year,” adding, “Weather, pests and diseases, and information on cultivation timing are needed throughout farming, so we thought that if we added farming information, it could become an app people reach for all year round.”

There is also a calculation aimed at meeting future customers in advance. Because farm machinery is expensive equipment purchased once every few years, it is important which brand comes to mind first at the time of purchase. Lee explained, “If even those who do not yet have Daedong farm machinery become familiar with the Daedong brand through the app, they will take another look at Daedong when they later buy farm machinery.”

Lee cited farming information functions useful even without farm machinery and active promotion as reasons for the increase in subscribers. Since 2025, Daedong has run online advertisements targeting users interested in farming, with AI Daedongi and the AI farm work journal at the forefront. At exhibitions and fairs, it held hands-on events where visitors could try the functions directly.

Among users’ cultivated crops, rice is the most common, followed by potatoes, sweet potatoes, peppers, and mandarins. There are also many perilla and apple farms. Lee said, “Among customers who own farm machinery, many are rice paddy farmers, while field crop farmers come in large numbers because of AI Daedongi and farming information.”
 

AI Daedongi receives 170,000 questions… most are on-site questions about sowing timing, pests and diseases, and more

At the center of the expansion is AI Daedongi. In the two years and six months since its launch, the number of questions received exceeded 170,000 as of October 2026. That is an average of about 5,700 per month. Most questions are about crop cultivation methods, pest and disease management, and how to use farm machinery. Lee said, “The most common cases are questions about things that have to be decided immediately, such as when to plant garlic, when to drain a rice paddy, or what supplementary fertilizer to apply and when,” adding, “Rather than asking in order to study, the vast majority are on-site questions people encounter directly while farming.”

The response has been particularly good from users with limited farming experience. In late 2025, a post appeared in the Daedong Connect community by a novice farmer who said he had moved to the countryside after retirement and started farming. He wrote that he was learning by planting potatoes, garlic, onions, and perilla, but that pest and disease control and the battle against weeds were the hardest, and that he hoped to farm enjoyably with help from the app going forward. There was also a post saying that many cucumbers were growing in a small garden but were not getting bigger, so the user asked AI Daedongi, which explained the cause and solution, and now they are growing well.

Lee said, “When farming, there are many cases where people have no one to ask right away, whether about pests and diseases or weeds, and end up wandering around searching on their own,” adding, “The starting point of AI Daedongi was to let those people ask right away anytime and receive help.” She added, “I feel the greatest sense of achievement when I confirm that it has actually helped someone’s farming.”
 

Photo and voice questions account for 40%… 84% of answers score 7 or higher for accuracy

Users’ voices have changed the functions as well. In the early days after launch, the only input method was typing text. About two months after launch, a request was posted in the community asking for photos of pests and diseases to be interpreted, and when the question records were reviewed, there were many cases in which it was difficult to give an accurate answer because symptoms were explained only in writing. There were also many users who found entering long text difficult. Daedong added voice question and photo question functions in the first quarter of 2025. Since then, among users who asked AI Daedongi questions, the share using voice and photo functions rose from 28% in the second quarter of 2025 to 37% in the second quarter of 2026. Based on all questions, questions entered in text account for 60%, while photo and voice questions account for 40%. Lee said, “After the photo and voice functions were added, use became quite active.”

AI Daedongi is based on a generative AI model from a leading global company. On top of that, Daedong has independently built reference materials for answers, including data on major crop pests and diseases and cultivation technologies secured in cooperation with the Rural Development Administration, as well as user guides and catalogs for all Daedong models. The company explains that the biggest difference from other agricultural apps is that users can ask about farming questions and how to use their own farm machinery in one place.

To improve answer accuracy, retrieval-augmented generation (RAG) technology was also applied. For questions related to farm machinery manuals or Rural Development Administration materials, the system searches the materials provided by the company and answers based on that content, reducing the room for AI to fabricate answers. It does not answer questions unrelated to agriculture. Lee said, “In farming, a single piece of incorrect information can change a year’s harvest,” adding, “We reflect the ‘like’ and ‘dislike’ marks users indicate for each answer in our checks, and we refine farm machinery answers in discussion with service field staff and specification-related answers with experts from the R&D department.”

Daedong measures answer accuracy through verification tests before reflecting new learning materials in the service. In the most recent verification in November 2025, it entered 396 questions that farmers would realistically ask, targeting 11 crops and nine Daedong farm machinery models. Each answer was scored out of 10 for accuracy, relevance, and satisfaction, using both human evaluation and AI automatic evaluation. As a result, accuracy averaged 8.5 points, and relevance to the question averaged 9.5 points. Answers that received 7 points or higher for accuracy accounted for 84%. However, these are the results of verification conducted internally by Daedong.

An AI farm work journal written by speaking, and a home screen with larger text for older users

The function Lee is most attached to is the AI farm work journal. She said, “Many people said that when they write a farm work journal by hand, they are tired after work and find it hard to write entries day by day, and that if they write them in a notebook, it is hard to find them later.” The AI farm work journal automatically creates a farm work journal with AI when users enter work details by text or speech, and stores them by date. The company explains that completed farm work journals can be downloaded and used as documents for Good Agricultural Practices (GAP) and eco-friendly certification or subsidy applications.

Because half of users are aged 55 or older, the screen is also being redesigned. In the previous version, the text was small and users had to tap several times to reach the desired function. In the recent home screen revamp, the text was enlarged and manuals were made easier to find. The procedures for registering farm machinery and farms were also reduced. Farm registration was changed from directly searching for an address to immediately designating a parcel. Some detailed function screens still remain in the previous version, so a full revamp is planned going forward.
 

Maintenance guidance before breakdowns and transmission of work plans… connected to AI tractors through data

Daedong Connect, which broadened its user base with farming information, is now upgrading farm machinery management, its original function. Currently, it only sends an alert when a fault code appears, but in the future it plans to analyze operating data coming from tractors and combines and notify users of maintenance timing before a breakdown occurs. For equipment that has been used for a long time, it will also provide guidance on replacement timing based on usage history. Lee said, “The goal is to reduce cases in which machines suddenly stop during busy farming seasons and to lessen the burden of repair costs.”

In the long term, the focus is on “connection,” linking farmland, workers, and farm machinery through data. In April 2026, Daedong launched the AI tractor “HX1400 AI” and implemented unmanned autonomous work in which, if a farm instructs work through the app, the tractor plans a route and performs the work on its own even without a person riding on it. Going a step further, the company plans to enable farms to set up a plan on the web or app by selecting the parcel to be worked, implement settings, and the worker, and then have that plan transmitted to the tractor. The next day, when the worker gets on the tractor, the work details will appear on the monitor, and results such as work time and fuel consumption by parcel will be sent back to the app.

Lee explained, “Even if the worker changes or is a beginner, the aim is for the work to proceed as the farm owner wants, and for farms to be able to handle everything from planning to results analysis themselves,” adding, “This is a function especially needed in North America, where farmland is large and there are many workers and farm machines.” Since 2025, Daedong has been operating a service of the same nature in North America and Europe under the name “KIOTI Connect.” 

10,000 monthly users… the task is to make it an app people use every day

Subscribers have increased quickly, but monthly users in 2026 average 10,000, about one-sixth of subscribers. That is why Lee cites steady usage as the biggest task. She said, “When I heard the news that we had surpassed 60,000, I felt responsibility before joy,” adding, “When there were few members, we could look at questions or community posts one by one, but once it became 60,000, not missing customer reactions became the most difficult thing.” Daedong has systematized a procedure of checking community posts and app errors every day, fixing issues with the development team when inconveniences are seen, and confirming whether they have been resolved.

Linking the farm work journal with public services is also a remaining task. Currently, the farm work journal does not directly link data with certification or subsidy-related agencies. Lee said, “Subsidies and certifications are well implemented in Nongsaro (the Rural Development Administration’s agricultural technology portal), so we have begun discussing ways to link with that side.”

Farm machinery location and operating data are collected and used based on consent to use personal information obtained during the dealership sales process. Daedong explained that it manages data by encrypting it in accordance with relevant laws and regulations, including the Personal Information Protection Act, and the company’s personal information processing policy. All services are free. Lee said, “Rather than generating profit through monetization, we are focusing on a direction that provides practical help to farms,” adding, “We will grow it into an app that solves everything from machinery management to farming concerns in one place through Daedong Connect, and that farmers turn to every day.”