In ZTE's Nanjing Binjiang base, more than 100,000 warehouse locations are arranged layer by layer in the three-dimensional warehouse. Four-way shuttles, elevators, stackers and other equipment work together to continuously distribute materials to the production line.
"Which box should be delivered to which location on the production line at what time depends entirely on multiple agents making decisions and giving instructions." Zhou Jianfeng, ZTE Vice President, said using the agent responsible for warehousing optimization as an example. When there are no picking tasks at night, it optimizes the arrangement according to the next day's production orders and moves frequently used materials to the nearest end in advance, reducing the monthly picking distance from more than 400,000 meters to 160,000 meters.
Today, there are more than 80 "digital employees" working on the business line in the base. Some arrange production plans, some monitor order circulation, and some are responsible for product life cycle management. The daily word consumption in the supply chain scenario alone exceeds 9 billion, and it continues to grow rapidly.
The exploration of a factory is the epitome of the acceleration of the implementation of intelligent agents.
From the beginning of the year, the open source intelligent OpenClaw started a "lobster farming" craze, to the intensive launch of intelligent products such as "Doubao Work", "Qianwen Office" and "Kimi Work", and then to the intelligent agents being "employed" to become employees in enterprises and research assistants in laboratories... In the first year of the "15th Five-Year Plan", artificial intelligence (AI) is accelerating from conversational products that can "talk" to intelligent agents that can "do things".
"Intelligent agents are an important form of artificial intelligence products and services." Chen Xi, a researcher at the National Development and Reform Commission's Macroeconomic Research Institute, believes that it is an intelligent system that uses large models as its capability base and has autonomous perception, memory, decision-making, interaction and execution capabilities. In short, the intelligent agent can think and plan independently and use various tools to help users get things done.
Able to think and work, various intelligent products and services are emerging at an accelerating pace and are being integrated into various economic and social fields.
"The cows are not eating, their body temperature is 39.5 degrees Celsius, and their milk production has dropped." In Mengniu's pasture, front-line operators describe the symptoms of the cows to the intelligent agent, and after two rounds of dialogue, they can get an accurate diagnosis and treatment plan.
This "AI doctor" was created by Wang Wangoyin, an employee of Mengniu Milk Source Division. Traveling to various pastures all year round, he found that when cattle in remote pastures had abnormal health, it was difficult to receive treatment immediately. Having never learned programming, he "fed" relevant authoritative works and years of work experience to Tencent's office intelligence WorkBuddy to develop an AI tool for precise diagnosis and treatment of cows. Now, the time for diagnosis and treatment in a single farm has been shortened from 1.5 days to 2 hours.
In the field of government affairs, intelligent agents are gradually embedded in the public service system. In Qiaokou District, Wuhan, after grid workers input residents' demands, the intelligent agent "Baidu Lazi" can generate standardized replies based on community policies and business rules, and complete the classification of appeals, and the generation and distribution of work orders; on the terminal side such as mobile phones, intelligent agents The body starts early verification, "Go home at 6 p.m. and book a car for me." As soon as the words are spoken, the Nubia smartphone equipped with Doubao Mobile Assistant immediately recognizes the time and destination, then calls the taxi service, selects the car model and route, without the user having to open the application layer by layer...
Having passed the proof-of-concept stage, my country's smart agent market is entering the productivity verification stage. According to survey data, the adoption rate of enterprise intelligence in my country has rapidly climbed from 17.3% at the end of 2024 to 40.3% in the first half of 2026.
"The agent transforms the large model into an agent for specific tasks through the closed loop of 'large model + memory system + tool calling + planning ability'," said Wu Hequan, an academician of the Chinese Academy of Engineering.
Why is intelligence booming? Technology, market and policy are three key words.
——Artificial intelligence technology continues to make breakthroughs. Zhong Xinlong, associate researcher at the Artificial Intelligence Research Center of CCID Research Institute, said that since this year, the basic model capabilities of DeepSeek, Zhipu, MiniMax and other manufacturers have continued to improve, and key technologies such as task understanding, planning, tool use, and group collaboration have made new progress. The "brain" is stronger and the "hands and feet" are more flexible, supporting the rapid popularization of intelligent agents.
——The release of market demand is accelerating. Relevant data shows that in the first half of 2026, the number of monthly active users of domestic AI native applications has reached 499 million, a year-on-year increase of 85.4%. Users' use of AI is rapidly penetrating into transaction processing scenarios. Chen Xi believes that the application of intelligent agents in repetitive and regular scenarios such as customer service and marketing can effectively reduce costs and increase efficiency. The acceleration of digital and intelligent transformation has led to the expansion of process automation to intelligent decision-making. Users' willingness to pay has increased. Various factors have promoted intelligent agents to open up incremental market space.
——The system design is continuously improved. This year's government work report mentioned "intelligent agents" for the first time; the three departments jointly issued the "Implementation Opinions on the Standardized Application and Innovative Development of Intelligent Agents", which clarified 19 typical application scenarios; a series of 7 national standards in the "Artificial Intelligence and Intelligent Agent Interconnection" series were released, effectively filling the gaps in standards in the field... Adhering to both innovative breakthroughs and standardized applications, the institutional framework for the development of intelligent agents has been accelerated.
At the same time, it should be noted that there are still many challenges in making "digital employees" better qualified for their jobs and more effective. The "KPMG 2026 Global Technology Report" shows that 88% of the companies surveyed have integrated intelligence into their systems, but only 24% of companies can achieve return on investment.
In Zhong Xinlong's view, the stability of intelligent agents in performing tasks for a long time is still lacking, and it is difficult to penetrate into core business links. Enterprise investments have not yet been generally realized as benefits. In addition, security risks such as privacy leaks, unauthorized operations, and out-of-control behavior also restrict the large-scale implementation of current intelligent agents.
Looking forward to the future, the "15th Five-Year Plan" outline outlines a new blueprint for the construction of Digital China, encouraging technological innovations such as intelligent agents, and comprehensively promoting the empowerment of digital intelligence technology; the "Opinions on the In-depth Implementation of the "Artificial Intelligence +" Action" issued by the State Council clarified that by 2030, the penetration rate of new generation intelligent terminals, intelligent agents and other applications will exceed 90%...
Multiple combined forces activate development momentum. As the underlying technology of artificial intelligence continues to evolve and relevant standards and rules are gradually improved, the application boundaries of intelligent agents will be further expanded, from local empowerment to global penetration, becoming an important force in reshaping production methods, lifestyles and governance methods.
