Double 11 E-commerce AI Shadow War: The Different Calculations of Qwen, Doubao, JD, and Pinduoduo

This Double 11, there is a special new participant in the shopping cart.

It can screen products, compare prices, stack coupons, and even proactively apply for price-protection refunds on behalf of consumers; it can also monitor inventory, analyze advertising performance, and generate marketing materials in bulk for merchants. The complex and cumbersome processes that once required repeated back-and-forth between buyers and sellers are now being taken over by AI.

On October 13, Alibaba launched its personal intelligent assistant "Qwen Go!", bringing shopping price protection, price comparison, and handling of daily-life matters into its AI capability circle. The next day, the Qwen Office e-commerce edition was released, attempting to use a single AI workbench to unify management of merchants' cross-platform businesses.

Almost at the same time, Doubao mobile assistant encountered a collective boycott.

On October 14, Doubao mobile assistant announced the latest progress of the Screen Automation Execution Protocol (SAEP), with more than 150 apps explicitly refusing to be operated automatically by AI. High-frequency apps such as Taobao, Alipay, Meituan, and Pinduoduo are all on the list, while JD.com has not yet appeared.

On one side, e-commerce platforms are racing to embed AI into the shopping chain; on the other, mainstream apps are firmly guarding against external AI interfering in their own transaction processes.

For Qwen, e-commerce is the best entry point to consumer-facing personal AI assistants and enterprise-level agents; for Doubao, shopping is the core scenario for validating cross-app task execution capabilities; and for JD.com and Pinduoduo, the value of AI has always revolved around traditional e-commerce metrics such as product selection, conversion, and operational efficiency.

Behind similar capabilities lie completely different business calculations. This new contest surrounding the shopping cart is fully emerging during Double 11.

01 Alibaba's Two Sets of Accounts

Alibaba's AI layout for this year's Double 11 unfolds simultaneously along both the user and merchant ends.

On the user side, Qwen's role is upgrading from "shopping advisor" to "shopping agent."

In May this year, the Qwen app was fully integrated with Taobao, accessing more than 4 billion products on Taotian, allowing users to complete the entire process of discovery, comparison, ordering, and after-sales through natural language. Taobao app also launched the Qwen AI shopping assistant internally, focusing on AI try-on and coupon coordination.

Taobao and Tmall's formats are more concentrated in shelves, where users need to simplify their needs into keywords for search and discover and compare desired products. Now, Qwen is trying to reconstruct the interaction logic. For example, when buying furniture for a small apartment, users only need to describe their living space and budget, and AI can automatically complete condition matching and screening.

The "Qwen Go!" tested on October 13 goes a step further, beginning to intervene in specific matters such as shopping price protection, flight compensation, and car maintenance inquiries. From "helping consumers find products" to "handling matters for consumers," Qwen is beginning to leap from an intelligent search box to the first entry point for users to express consumption needs.

This Double 11 naturally becomes the examination room for Taotian's new "shopping agent," with assessment metrics being order conversion rate, average order value, and repurchase rate. During Double 11, rules are complicated, and AI calculating the optimal solution for users can directly reduce the cost of price calculation and recover orders abandoned due to complexity.

On the merchant side, the Qwen Office e-commerce edition attempts to replicate this logic in the enterprise services field.

The Qwen Office e-commerce edition released on October 14 provides functions such as cross-platform business dashboards, trend insights, material generation, and cross-platform ROI optimization, and connects with business tools such as Shengyi Canmou, Qianniu, and Youzan. Data from the launch event shows that e-commerce industry users already account for more than 20% of its paying users.

Multi-platform operation is the norm for merchants today, but order, inventory, and advertising data are scattered across different backends. Qwen Office attempts to become an independent third-party e-commerce agent, unifying scattered data into the same AI workbench.

The ambition of this positioning lies in the fact that if merchants become accustomed to handing reports, marketing plans, and inventory strategies from various platforms to Qwen Office, the business context and workflows it accumulates will translate into an extremely solid B-end software moat.

Although at this stage Qwen Office's core capabilities remain in data integration and decision assistance, with specific ad placements still limited by API permissions and manual confirmation, its strategic intent is already very clear.

02 Whose Wall Did Doubao Hit?

If Qwen's shopping capabilities are built within Alibaba's own ecosystem, then what Doubao mobile assistant is trying to solve is another ultimate problem.

Can users skip the app and directly assign tasks to AI?

In September this year, relying on mobile system permissions, API interfaces, and GUI (graphical user interface) automation technology, the Nubia Doubao second-generation mobile assistant began trying to understand user instructions, simulate manual clicks, swipes, and input, and execute question search, price comparison, and ordering across apps.

This GUI automation model greatly lowers the threshold for invoking existing apps, and also pushes agents to the forefront of ecosystem confrontation.

Although simulating manual operations can save users cumbersome steps, it easily touches platform red lines regarding account permissions, payment security, and risk control. The deeper conflict lies in the redistribution of commercial traffic entry points.

Traditional e-commerce business models are built on the chain of "search—recommendation—browsing—advertising," with platforms monetizing through exposure and traffic allocation. Once consumers fully delegate their needs to external AI, product screening and comparison will be completed outside the app.

When external AI reorders products and distributes traffic, e-commerce platforms' original advertising monetization mechanisms and recommendation algorithms face the risk of becoming ineffective. Platforms may still provide payment and fulfillment, but they lose their most core user attention and traffic control. This is precisely the deeper reason why Doubao mobile assistant suffered a collective boycott.

As mentioned earlier, as of October 14, the list of those refusing Doubao's SAEP protocol has increased to more than 150, including giants such as Taobao, Alipay, Meituan, Pinduoduo, and WeChat.

What needs to be clarified is that what platforms resist is external assistants' automated unauthorized operations on their own apps, not the technology itself. Taobao has its own AI assistant, Pinduoduo has in-app AI search, and each welcomes AI to improve transaction efficiency on its own platform, but firmly guards against external agents crossing the rules to erode user relationships.

For Doubao, shopping is the touchstone for proving its own cross-app agent orchestration capability; but for transaction platforms, once this capability scales, the app will be reduced to a pure back-end fulfillment tool. The more AI wants to become an independent super entry point, the greater the ecosystem resistance it faces.

03 JD.com and Pinduoduo, another transaction account

On Doubao's rejection list, JD.com is a subtle exception.

As of October 14, JD.com-affiliated apps were not explicitly included on the rejection list. In commercial logic, this implies a possible choice: if external AI can bring new orders while transaction fulfillment remains in its own hands, the platform does not reject moderate openness.

Xiaohongshu, once jokingly called the great open loop by the industry, repeatedly opened and shut down external link redirects in the process of cultivating a platform e-commerce ecosystem. Previously, Xiaohongshu launched the "seeding direct" feature, allowing advertising notes to directly redirect to external e-commerce platform pages such as Taobao and JD.com. This approach of giving up part of the front-end decision traffic in exchange for orders and conversion revenue across the entire network is precisely the compromise product after the game between traditional traffic platforms and fulfillment platforms.

JD.com's long-accumulated self-operated supply chain and logistics fulfillment system is its most solid line of defense. Even if users compare prices through external AI, what ultimately determines the transaction is still price, inventory, and delivery timeliness. External AI is responsible for stimulating and screening demand, and JD.com undertakes fulfillment; the logic remains smooth.

Of course, JD.com is also building its own AI defense line. On the user side, JD.com upgraded its AI shopping assistant "Dongdong" and launched the JoyAI real-time video digital human, connecting shopping and life services; on the merchant side, it launched the Jingmai AI operation center, planning to provide digital human livestreaming services to more than 90,000 merchants during Double 11.

Compared with JD.com's high-profile display, Pinduoduo's approach is more pragmatic and more concealed.

In May this year, Pinduoduo quietly launched an AI search feature. When users input complex long-tail demands such as "find low prices, limited brands, black labels, or tens of billions in subsidies," AI precisely breaks down the conditions and matches them against the product library. Before that, the platform had also successively launched C-end features such as fitting rooms, AI friend interaction, and AI interactive dramas, either to boost activity or to assist decision-making and conversion.

It can be seen that Pinduoduo positions AI purely as a high-efficiency internal tool. Whether it is product distribution or activity boosting and retention, the launch of a series of small features all carries a clear purpose, while not overly interfering with users' habitual shopping behavior. This forms a sharp contrast with the industry practice of starting from scratch to build conversational "shopping guides" and attempting to change users' shopping habits.

There are rumors that Pinduoduo internally still "follows" ancient-style programming, because the group does not care whether employees use Vibe Coding to improve efficiency, but cares more about actual output. This extreme efficiency orientation is exactly the same as when current Pinduoduo co-chairman Zhao Jiazhen built the agricultural product upward supply chain in his early years.

On the merchant side, Pinduoduo's official open platform has long opened order and logistics interfaces to merchants, allowing third-party service providers to develop AI customer service and product management tools under authorization. A Pinduoduo merchant said that platform staff even encourage the use of ByteDance-affiliated e-commerce material generation tools and will teach them hand by hand.

Help merchants manage business, make decisions for users. This Double 11, the competition in e-commerce AI has shifted to the "agent layer."

But no matter how the front-end entry point evolves, product supply capability, fulfillment timeliness, and after-sales protection remain the cornerstone determining whether a transaction can be established. It can change the direction of traffic flow, but it cannot overturn the most basic laws of supply and demand.