Text by | Peng Qian
Edited by | Qiao Qian, Yang Xuan
At the beginning of the new fiscal year on April 1st, the Alibaba China E-commerce Business Group led by Jiang Fan is planning growth points that can be achieved in combination with AI.
According to 36Kr, Taotian Group’s current exploration of AI will focus on the AI to B direction. Its core OKR is the retention rate of AI tools on the merchant side and the GMV growth brought by AI. Against the backdrop that in 2026, Open Claw triggered the explosion of AI Agents, Alibaba CEO Wu Yongming proposed to establish a new Alibaba Token Hub (ATH) Business Group, and all related businesses need to be commercialized around Tokens.
Organizational changes have also taken place accordingly. According to 36Kr, Zhang Kaifu, the person in charge of the AI business of the China E-commerce Business Group, is no longer in charge. The “Intelligent Search and Recommendation Products” division, which undertakes the AI business, has been adjusted into two departments: “Platform Users and Products” and “Intelligent Algorithms”. The “Future Innovation Division” responsible for multi-modalities has been integrated into the ATH Business Group.
“In the new year, ATH will lead the group’s AI strategy. There’s no need for other businesses to reinvent the wheel,” a source close to Taotian Group told 36Kr.
At the beginning of last year, the China E-commerce Business Group established the Core Search and Recommendation Intelligent Products Division, which was led by Zhang Kaifu. In the past year, his main work was to integrate multiple teams such as search and recommendation algorithms, user algorithms, merchant algorithms, and creation algorithms to improve the accuracy and efficiency of the search and recommendation systems. He focused on exploring the AI to C aspect, and the AI penetration rate at the user level was his core OKR. This required continuous optimization of the user experience. As a result, Taobao launched six products such as “AI Universal Search”, “Help Me Choose” (AI shopping guide), and “Scan to Search for Low Prices” (AI search for low prices). In the AI to B aspect, AI was used to enhance advertising effectiveness and support merchant growth.
His last public appearance was at the 2025 Tmall Double 11 kick-off meeting, where he delivered a speech titled “AI Reconstructs Taobao” and revealed that the platform used AI to understand and organize 2 billion products more accurately, achieving double-digit growth in traffic matching efficiency: including a 20% increase in search relevance under complex semantics, a 10% increase in the click-through rate of the recommendation information stream, and a 12% increase in the ROI of merchant advertising.
In Q2 of last year, Jiang Fan also said at the earnings conference that Taotian was committed to using AI to improve the user experience and reconstruct the search advertising system. This was Taotian’s priority, and some significant results had been seen. According to 36Kr, the Search and Recommendation Division did optimize the search strategy. For example, it modified the logic of search and recommendation in combination with AI technology to increase the proportion of search conversions into transactions, which helped Taotian increase the order volume to some extent.
However, compared with AI to C, AI to B obviously has a clearer commercialization path. This is also an important reason why Anthropic’s annualized revenue soared to $30 billion recently, surpassing OpenAI. With a large number of merchant resources, it has become a necessity for Alibaba’s e-commerce business to make a new design around the Token economy.
A person close to Taotian Group told 36Kr that Taotian is currently planning to recruit “Fourth-generation service providers”, that is, AI Agent-type service providers. The first three generations were human operation agents, software customization, and SaaS respectively. In terms of specific actions, the Qianniu platform originally launched for merchants will also be upgraded to “Qianniu Claw”. Taotian will officially promote this new product around June 18th, and introduce fourth-generation service providers through it, making them its small Claws or Skills. An Alibaba employee said that currently, merchants’ core demands on the platform mainly fall into three aspects: increasing traffic subsidies, improving operational efficiency, and reducing product selection costs. The core function and goal of Qianniu Claw are to reduce the operating costs of various merchants and improve operational efficiency.
According to Taotian’s vision, Qianniu Claw can provide customized services for big brands and ready-to-use services for small and medium-sized brands. At the same time, it will deploy merchants’ business systems, applications, and data resources on Alibaba Cloud. In this model, when merchants use Qianniu Claw, they will consume Tokens, and the fees will be borne by the merchants alone or jointly by the merchants and service providers.
This change is particularly beneficial to small and medium-sized sellers. According to the feedback from many merchants, opening a Tmall store is quite complicated. It not only requires sufficient manpower but also the search for a suitable operation agent. With Qianniu Claw, all these can be solved by AI Agents. The e-commerce departments of big brands often have hundreds or even thousands of people, and they can also use Qianniu Claw to reduce staff, and some functions can be Agentized.
Theoretically, this will be a win-win situation. It can help merchants reduce operating costs, and at the same time implement Alibaba’s Token economy on the e-commerce side. In the future, the measurement index of e-commerce growth will no longer be just GMV. This is also due to the pressure of revenue growth. Alibaba’s latest Q4 2025 earnings report shows that Taotian Group’s revenue growth rate was 6%, and the year-on-year growth rate of CMR (Customer Management Revenue) dropped significantly to 1%.
According to the above source, Qianniu Claw is expected to increase revenue from two dimensions: on the one hand, if merchants significantly reduce operating costs and increase gross profit, they will be more willing to increase their advertising budgets; on the other hand, instead of calculating based on advertising fees, it will be based on Token consumption to ensure the overall revenue growth of the group.
At the recent earnings conference, Wu Yongming gave the guidance for AI revenue for the first time: within the next five years, the revenue from AI and cloud should reach $100 billion annually. According to the above person close to Alibaba, according to the new Token consumption algorithm, this $100 billion can also include the AI revenue generated by the e-commerce business.
On the merchant side, Taotian Group is also upgrading the “Merchant Operation Index” and exploring full AI-based operations. This means that AI will also deeply penetrate into the operation level, allowing AI to act as an inspector instead of humans. For example, it can determine whether a merchant’s behavior requires a fine or whether a product should be taken off the shelves, and make corresponding adjustments accordingly.
Based on this, a Taotian employee told 36Kr that some of Taotian’s products and technologies are also being reorganized according to AI agents. For example, in the past, products were directly connected to traditional business requirements, but now products will provide skills for Agents. Resources will be tilted towards demands such as intelligent AI refund robots and intelligent fines.
This year, the implementation of AI-based office work in Alibaba’s e-commerce will be more thorough. The e-commerce business has given employees OKR indicators related to AI, such as the penetration rate of AI tools and the problem order resolution rate; the international e-commerce business requires all employees to learn, punch in, and take exams to win points. Currently, the points are not linked to performance, but there are additional incentives. This kind of exam is similar to the Business Conduct Guidelines Certification Exam and Data Security Compliance Certification Exam held within Alibaba every year. It is not directly related to the evaluation of professional skills, but it clarifies the importance of AI from a institutional perspective.
Management hopes that employees will use AI products more frequently to increase work efficiency. An employee of the international e-commerce business told 36Kr that they use an AI office software called Qoderwork, similar to Wukong, which is equivalent to an AI assistant. “Two years ago, the only AI exploration was in customer service. AI was the first to handle merchant consultations, and the responses were quite stupid at that time. Later, AI gained the ability to review products, and merchants could apply through AI, which reduced a lot of trivial matters.”
According to 36Kr, since March 25th, e-commerce employees can also queue up to get Tokens and use them in their work processes. In addition, e-commerce business personnel are also actively applying to use the Wukong system.
In terms of model selection, Taotian E-commerce is currently strongly bound to the Qwen large model. This is not only to support products within the Alibaba ecosystem but also involves important issues of business data privacy and security. According to 36Kr, last year, Taotian employees could choose from Qwen, Deepseek, and Kimi when selecting AI tools.
Under the group’s All in AI strategy, it took Alibaba two full years to transition from GMV to Tokens. Alibaba is still continuously undergoing self-transformation to secure a ticket to the AI Agent era.
