Original · Unique Research · 2025-10-31
Editor’s note: This complete English edition preserves the original report’s first-person commentary and the interviewee’s claims. Farm-area, profitability, growth and performance figures are source-attributed, not independently audited results. The Chinese land-area unit mu is retained. No publication location is inferred. ChunYun Smart Agriculture is an editorial English rendering of the Chinese company name, not a verified official English legal name.
Have you noticed something? When we talk about AI, we always seem to talk about large models, video generation, and intelligent customer service—but almost no one talks about farmland. We discuss technology with great enthusiasm, yet the actual land where our food is grown has somehow been left out of the conversation. It is almost as if anyone who still farms is a little ‘behind the times.’ But is that really true?
In this interview with Xu Lianyun, founder of ChunYun Smart Agriculture, we happened to discuss how AI is being put to work in agriculture. You might think this is not a hot sector: it is not glamorous, it is not a high-frequency use case, and it is not even an easy story to tell investors. Yet for me, this conversation was especially illuminating.
Here is someone who may look like the most ‘down-to-earth’ person in AI, building the most literally grounded AI business—and who may genuinely change something.
Source-photo note: The original article includes a conference photograph with the backdrop “China (Guangxi)–ASEAN Artificial Intelligence Enterprise Conference” (CAAIEC). The illustrative photograph is not reproduced in this text-first edition.
Who in the AI industry understands agriculture best?
I cannot say for sure.
Ask most entrepreneurs in the AI industry and agriculture probably has little or nothing to do with what they do. Xu Lianyun is different. He previously worked on commercialization at a leading global agricultural technology company and understands how agriculture actually works. He is not someone who ‘spotted a trend and pivoted into agriculture.’ He came from this land and chose to return to it.
He calls theirs the most ‘down-to-earth’ AI company. I see it differently: they may be among the people who understand agriculture best in the AI industry, and who understand AI best in the agricultural industry. Rarer still, he can bring these two seemingly opposing fields together and turn them into a credible direction.
Many people now talk about ‘cross-industry integration,’ but frankly, it cannot be done without real expertise. You need to understand the rhythms of agriculture as well as the logic of models. You need to know when a field should be fertilized and when an agent should be triggered to make a decision. Combining these bodies of knowledge is difficult in itself, let alone turning that combination into a product.
Why does farmland need AI?
Because there are not enough people.
This was the first pivotal shift Xu Lianyun identified: people no longer want to farm. A shrinking pool of willing, knowledgeable farmers is the real difficulty confronting agriculture today. China has 1.4 billion mu of staple-crop farmland, while the ASEAN countries together have 600 million mu. This is not a matter of experimenting on a few plots; it is a systems-engineering challenge.
Large farms have become increasingly common, especially following transfers of land-use rights. The area of farms above 1,000 mu is growing by 15% each year. In theory, greater scale should mean higher efficiency and stronger profitability. In practice, the reverse is true: farms of 1,000–3,000 mu still have a reasonable profitability rate, but it drops sharply for farms of 3,000–5,000 mu, and most farms above 5,000 mu do not make money.
Why?
Because they are too large to manage. However capable one person or one team may be, they cannot keep an eye on thousands of mu. They do not know which plot has developed a problem, where pests have appeared, when water should be released or drained, or even where replanting may be needed. That is where AI creates value—not by replacing people, but by making up for a shortage of human capacity.
How do you give farmland a ‘physical examination’?
With sensing systems plus large-model analysis.
Xu Lianyun says farmland is like the human body: it also needs ‘health management.’ We undergo medical examinations that check liver function, blood lipids, and tumor markers, so why should farmland not receive comparable care? Pest damage can be modeled as an ‘infection,’ weeds as ‘abnormal proliferation,’ and drought as ‘dehydration.’ All of it can be modeled.
So how can these problems be detected?
Smart-agriculture technology has moved well beyond the old stage of ‘installing a few cameras and calling it done.’ Satellite remote sensing, low-altitude drone imaging, multispectral and hyperspectral imagery, and all kinds of ground-based sensors together form farmland’s ‘sensory system.’ In the past, however, everyone built only ‘one organ’: one company built the eyes, another the ears, and another the nose. No one built the entire ‘nervous system’ and ‘brain.’
What ChunYun Smart Agriculture does is integrate all of that sensory data and hand it to AI models for analysis and decision-making. It is like the final diagnosis and treatment recommendation a physician writes after reviewing a medical examination report.
Xu Lianyun says what they provide is an ‘electronic prescription for farmland.’
Once the diagnosis is complete, what comes next?
Make it understandable and actionable for the farm operator.
We often talk about data-driven decision-making, but have you ever met a farm operator who spends every day reading dashboards? Whether data is useful does not depend on how elaborate the charts look. It comes down to one question: can it help me solve a real problem?
That is exactly what ChunYun Smart Agriculture’s AI system does. It tells you where the problem is, recommends a solution, and produces an electronic report. Small farms can purchase the service one report at a time, much like going to a hospital for a checkup; medium-sized farms can buy a subscription; and very large farms can commission a customized deployment. Different farm sizes and budgets correspond to different service models.
Does that sound ordinary? In this industry, even taking a photograph of a field and transmitting the data can be difficult, never mind modeling, analysis, and intelligent reasoning. For many farm operators, the arrival of an ‘AI health-check report’ may mark the first time they can actually tell whether their land is healthy.
Is agriculture a public service or a business?
Neither. It is a responsibility.
Agriculture is not an industry that immediately excites most people. It cannot create special effects like video AI, nor can it lift conversion rates like e-commerce AI. But listen closely to Xu Lianyun and you will hear an unusually strong sense of purpose.
He said that when he previously worked on commercializing agricultural technology, customers did not treat his team as a conventional contractor. Instead, they saw them as people who had come to help. The services they provided genuinely addressed farmers’ pain points, and even government officials felt that ‘you are doing work that truly matters.’ That feeling of being needed was powerful enough to make him give everything else up and start another company.
I strongly agree with something he said: ‘If agriculture cannot make money, subsidies alone cannot keep it alive.’
Think about it. On one hand, fewer and fewer people are farming; on the other, food prices cannot simply be allowed to rise. So what will sustain food security—subsidies, or policy? Ultimately, farmers must be able to earn more. This is not a sentimental argument. It is the logic of survival.
Xu Lianyun and his team want to use AI to reduce management costs, improve crop health, and cut the use of chemical fertilizers and pesticides, creating a virtuous cycle.
One final thought.
Not every AI product needs to chase hundreds of millions of users. Some AI systems serve just one tract of land—but that land may feed hundreds of people.
Nor does every startup have to pursue the most glamorous story. Some fields will inevitably involve hard, exhausting work. But succeed in them, and you can make the world a little more stable.
That is the work ChunYun Smart Agriculture is doing, and that is the kind of person Xu Lianyun is. There are no florid phrases from a PPT deck and no résumé built on job-hopping among tech giants. He is simply doing what he knows, moving agriculture forward one step at a time.
And I believe this kind of AI—the kind that may look the most ‘down-to-earth’—is the AI that truly stands in the soil and cultivates the future.
What do you think?
Interview Highlights: Q&A
Market and Opportunity
Q1: Why are you interested in the Southeast Asian (ASEAN) market?
Xu Lianyun: The market has enormous potential. We focus on staple crops—rice, wheat, and corn. China has nearly 1.4 billion mu of staple-crop farmland, while the ASEAN countries have more than 600 million mu. I also conducted an assessment during this visit, and we will consider entering the market in the future.
Q2: What challenges does the high-potential Southeast Asian market currently face?
Xu Lianyun: The challenge is that ASEAN countries have not progressed as quickly in developing large-scale agriculture. Farming is still dominated by smallholders working a few mu or a few dozen mu. Our technology, by contrast, serves large expanses of farmland: the larger the farm, the harder it is to manage and the more it needs information technology (IT).
Q3: Compared with Southeast Asia, how far has large-scale agriculture developed in China?
Xu Lianyun: China is moving from small-scale to large-scale agriculture with great determination and speed. Through transfers of land-use rights, farmland operations above 1,000 mu are growing by 15% every year. China now has tens of thousands of large farms of this kind, and these large farms are our principal customer group.
Industry Pain Points
Q4: Why did you choose agriculture—this ‘narrow gate’—for your second startup?
Xu Lianyun: I have a background in agricultural technology and a real passion for the industry. At my previous company, I derived a tremendous sense of purpose from feeling that I was genuinely helping some of the hardest-working people in China—farmers—and addressing the difficulty of farming. That commitment and sense of value drove me to start a second company.
Q5: What do you see as agriculture’s central pain point?
Xu Lianyun: The central pain point is ‘management.’ ‘Uncrewed operations’—such as drone seeding and spraying or autonomous farm machinery—are already highly mature because people no longer want to farm. But ‘operations’ are not the same as ‘management.’
Q6: Why is ‘management’ a more fundamental pain point than ‘operations’?
Xu Lianyun: Because the farms are simply too large to manage. Let me share one data point: in theory, the larger the farm and the more efficient the mechanization, the more profitable it should be. In reality, nearly 40% of farms covering 1,000–3,000 mu are profitable, while only a little over 10% of farms above 5,000 mu make a profit.
Q7: Why, then, are larger farms less profitable?
Xu Lianyun: The reason is simple: the farmland is too extensive for the operator to manage. The operator cannot promptly identify where weeds or pests and diseases have appeared, whether the field is level, when to release water, or where replanting is needed... If problems cannot be detected in time, management decisions cannot be made in time either.
The Solution
Q8: How does your product solve the problem of farms being ‘too large to manage’?
Xu Lianyun: Much of smart agriculture in the past consisted of concept demonstrations for To G, or government, projects. We are To B and focused on value. Farm operators do not need isolated data points, such as readings from a single sensor or camera. They need ‘full-view’ management decisions.
Q9: How do you provide those ‘management decisions,’ and what is the technological core?
Xu Lianyun: Our core technologies are large models and Agent systems. We integrate various kinds of hardware—satellite remote sensing, drones, sensors, and more—into an ecosystem that captures three-dimensional sensory information. Our brain, the model, then compares and analyzes that information against the scientific standards expected for the crop before automatically generating a conclusion and issuing an ‘electronic prescription.’
Q10: How do you charge farms of different sizes?
Xu Lianyun: We provide tiered services:
Small farms (1,000 to 10,000 mu): We offer a ‘service model’ and issue reports on a per-use basis. It resembles frequent health screening; in a rice field, the leaf age may need to be checked every 3 days.
Medium-sized farms (a few tens of thousands of mu): We provide a standardized MaaS (Model as a Service) product on a subscription basis.
Very large farms: We provide customized local deployment.
Social Value
Q11: What social value can this solution ultimately create?
Xu Lianyun: It creates two core forms of value:
Food security: Only by improving farmers’ ability to make money can we create a virtuous cycle, reduce the burden on the state, and address food security at its root.
Green agriculture: When we can make precise decisions and issue precise prescriptions, we can apply chemical fertilizers and pesticides accurately and in smaller quantities, sharply reducing the use of agricultural inputs. This advances green agriculture and also indirectly addresses public concerns about prepared dishes and food-ingredient safety.