China: The ‘Year of Popularization’ for AI Shopping

Marcus Zhan , Director of GMA… I will explain what I have read in Chinese Social Media. The ‘Year of Popularization’ for AI and make an article.

Shopping: From Tool to Exclusive Agent, Accelerating Implementation, and the
Changing Consumer Gateway”….

I have noticed that during the 2026 Spring Festival, major internet companies invested over 8 billion yuan,
using red envelope subsidies and other forms to complete a “user acquisition experiment” for AI applications, embedding AI into the entire consumption process.

AI shopping is currently entering an acceleration phase, driven by technological
advancements and ecological integration, showing trends of gateway restructuring, full-
scenario integration, and industry standardization. It is now at a critical stage of diffusion
to the mass market, poised to reshape the logic of e-commerce and consumption. The
2026 Spring Festival may become a special node, marking AI’s evolution from a
“function” to a shopping “infrastructure.”
Reported by Chen Ting and Li Yutong from the National Business Daily, Edited by Yu
Tingting

Iterative upgrades in the consumer sector converged during the 2026 Spring Festival.
This Spring Festival, major internet companies like ByteDance, Alibaba, Tencent, and
Baidu invested a cumulative total of over 8 billion yuan. Through forms such as red
envelope subsidies, scenario-based freebies, and Spring Festival Gala interactions, they
completed a large-scale “user acquisition experiment” for AI applications aimed at the
entire population.

AI assistants

When people struggled with choosing gifts for visiting friends and relatives, AI assistants
could already generate tailored recommendations in seconds based on the recipient’s
age, interests, and relationship closeness. From stocking up on New Year’s goods to
instant consumption, AI is embedding itself into the entire consumption process of
Chinese people in a subtle, “moistening rain” manner.

200 million orders WTF

Data shows that during the 2026 Spring Festival holiday, Alibaba’s Qianwen App helped
users place nearly 200 million orders with “one-sentence commands.” Tencent’s
Yuanbao had over 50 million daily active users at its peak. ByteDance’s Doubao set a
record on New Year’s Eve with 1.9 billion total AI interactions.
This is a profound transformation reshaping the underlying logic of consumption. Around
the “3.15” period of 2026, the National Business Daily’s Future Business Think Tank
focused on interviewing key players about the前沿 trends in “AI + shopping.” From the
multiple dimensions of current implementation status, evolutionary direction, and
industrial impact, we systematically analyze the future business landscape of this AI-
driven consumption revolution.

1. AI Shopping Implementation Accelerates
   Currently, AI shopping has entered an acceleration phase of practical application. Chen
   Yixin, Managing Director of CIC Consulting, told NBD reporters that industry
   development is primarily driven by a combination of technological advancement and
   ecological integration:
    Technology-Driven: The rise of cross-modal technologies (integrating text,
   images, voice, etc.) enables AI to better understand products, user needs, and
   shopping scenarios, laying the foundation for more intelligent guidance and
   interaction. Simultaneously, the rapid iteration of large model technology,
   especially improvements in Chinese language understanding and multi-modal

capabilities, provides the core engine for AI shopping applications.

Ecological Integration: The development of domestic AI shopping functions
deeply relies on the existing business ecosystems of major internet companies,
using AI to reconstruct and enhance their original e-commerce, content, and
service ecosystems.

“Compared with foreign counterparts, Chen Yixin stated that domestic AI assistants place
greater emphasis on “scenario depth” and “ecological synergy,” prioritizing closed-
loop implementation within their own advantageous e-commerce and content
ecosystems to pursue immediate improvements in user experience and business
efficiency. The competition focus is on penetration rates in vertical scenarios and
commercialization efficiency.“

Starting from typical cases, application scenarios, and development trends, NBD
summarizes the current main implementation status of AI shopping.
(A) Typical Cases of AI Shopping

1. Alibaba: Qianwen App + Alibaba’s自有 Ecosystem
   On January 15, 2026, the Qianwen App announced its full integration with
   Alibaba ecosystem services such as Taobao, Alipay, Taobao Flash Sale, Fliggy,
   and AutoNavi. It globally achieved AI shopping functions like ordering food,
   buying goods, and booking air tickets for the first time, opening testing to all
   users.
   (Image Source: Screenshot from Sina Weibo @Qianwen)
   After integrating with Taobao, the Qianwen App helps users solve the dilemmas
   of “what to buy, how to choose, and whether it’s worth buying.” Moreover, it not
   only provides shopping suggestions based on users’ real needs but also
   intelligently generates specific product recommendations based on Taobao’s vast
   product database and review system. Its core characteristic lies in Qianwen App’s
   choice to create a complete closed loop from “consumption planning” to
   “transaction payment” within Alibaba’s own ecosystem. While avoiding cross-
   platform blocking risks and ensuring service fluency, this model also deeply ties
   its development to the richness and openness of the Alibaba ecosystem.
2. ByteDance: “Doubao + Douyin E-commerce” Connects the Consumption
   Chain, “Doubao Mobile Assistant” Creates a System-Level Consumption
   Gateway
   In October 2025, ByteDance’s Doubao large model began embedding Douyin
   Mall’s transaction links. This enabled users to express needs through natural
   language, Doubao completed intelligent product matching, and then users could
   一键 jump to the Douyin App from the Doubao interface to place orders on
   Douyin Mall, forming a closed loop. Subsequently, Doubao iterated and
   upgraded: when users inquired about consumption needs, it could simultaneously
   match Douyin bloggers’种草 short videos and integrate Douyin’s local life group-
   buying services, integrating “consumption + content” within Doubao.
   Its core advantage lies in relying on ByteDance’s content e-commerce ecosystem
   to串联 AI recommendation and short video种草 conversion. However, the
   shopping function is currently in a relatively early stage; deep thinking mode
   cannot trigger shopping functions, and its adaptability to complex consumption
   needs is limited. Furthermore, the transaction and payment环节 still require a
   jump to Douyin Mall, failing to achieve a closed loop within the software.
   (Image Source: Photo by NBD reporter Li Yutong)
   In December 2025, ByteDance further released a technical preview version of the
   “Doubao Mobile Assistant,” collaborating with ZTE Nubia to launch the M153

engineering machine. Doubao Phone can complete cross-app price comparisons,
content screening, and quick order placement, shortening the path from demand
to transaction and achieving system-level deep AI integration. However, limited
by sensitive operations and ecosystems not yet fully connected, operations still
encounter obstacles requiring manual user confirmation, and a seamless closed
loop has not yet been achieved.

3. Moonshot AI (Kimi): Neutral AI Shopping Guide
   In the second half of 2025, Moonshot AI’s智能助手 Kimi added an AI shopping
   guide function. Leveraging its core advantage of long-context processing, it
   integrated product jump links from mainstream e-commerce platforms like
   Taobao and JD.com.
   Its core competitiveness lies in not being tied to its own e-commerce ecosystem,
   providing consumption decision services as a neutral third party. However, Kimi
   lacks its own e-commerce infrastructure; transaction links must pass through
   third-party intermediate pages, resulting in conversion efficiency far lower than
   the closed-loop ecosystem model. Additionally, its recommended links primarily
   point to third-party non-official stores, posing certain consumption risks.
   (B) Multi-Scenario Applications Front and Back End: AI Shopping Technology
   Implementation
4. Front-End Consumer Scenario Applications
   o Full-Link Intelligent Shopping Assistant: Represented by Qianwen and
   Doubao, evolving from a single product recommendation tool into an
   intelligent hub covering demand analysis, decision assistance, and
   transaction conversion. It has become a core consumer decision gateway
   for users.
   o Multi-Modal Visual Shopping Upgrade: Since the second half of 2025,
   breakthroughs have been made in AR virtual try-on, try-out, and photo
   search technologies. Douyin E-commerce upgraded its dynamic AR fitting
   room to support full-body dynamic outfit effect displays. Taobao’s photo
   search accuracy rate increased to 98.2%, significantly reducing the trial-
   and-error cost of online consumption.
   o AI Digital Human Interaction: Digital humans have evolved from fixed-
   script live streaming tools into intelligent anchors capable of real-time
   interaction, multi-language adaptation, and dynamic script adjustment.
   (Image: Luo Yonghao explaining a product, left is digital human live
   stream, right is真人 live stream. Source: Live stream screenshot)
   During the “618” period in 2025, the digital human Luo Yonghao
   completed its带货 debut. This digital human live stream attracted over 13
   million views, achieving a GMV exceeding 55 million yuan. Liu Yaping,
   Vice President of Friends and Head of Information, told NBD reporters
   that the next step for the company is to create a live streaming e-
   commerce AI agent, deepen digital human live streaming applications,
   and promote industry openness and standardization. Among these, the
   live streaming e-commerce AI agent will focus on building a专用 large
   model memory system based on industry professional knowledge,
   constructing an AI agent matrix covering core scenarios like digital human
   live streaming and intelligent customer service, achieving highly AI-driven
   business processes.
   o AI Intelligent Customer Service: Based on large models, intelligent
   customer service has evolved from simple Q&A tools into full-scenario

AI Agents capable of handling complaints, order modifications, after-sales
rights protection, and consumption recommendations, significantly
shortening response times.

2. Back-End Industry Scenario Applications
   o AI Empowered Full-Link Content Creation: AI can generate product
   detail pages, live streaming scripts, short video content, and marketing
   copy with one click. During the 2025 “Double 11” period, ByteDance’s
   collaborative office software Feishu announced its entry into “Double 11.”
   In its flagship function “Multi-dimensional Tables,” users can directly调用
   AI tools like DeepSeek for copy generation, Jimeng for image generation,
   and Kimi for image information extraction when creating new columns,
   integrating AI capabilities directly into the spreadsheet. The system can
   function in multiple scenarios and links such as complaint reviews,
   customer profile insights, product selection, product image generation,达
   人 management system, dealer management, and integrating live
   streaming data.
3. Three Major Trends in AI Shopping Implementation: Gateway Restructuring,
   Full-Scenario Integration, and Industry Standardization
   o Gateway Restructuring Trend: Evolving from a tool-type application
   within an App to a cross-platform, full-scenario personal exclusive
   consumption agent. Everyone will have their own exclusive AI shopping
   butler, enabling unified management of consumption needs across all
   channels. AI will become the core traffic gateway in the consumer sector.
   o Full-Scenario Integration Trend: Deep integration of online and offline
   consumption scenarios. AI connects online e-commerce, offline stores,
   instant retail, and local life services across all scenarios, breaking the
   boundaries between online and offline consumption.
   o Industry Standardization Trend: With the large-scale popularization of
   AI shopping, relevant compliance and regulatory systems will gradually
   improve. Rules regarding data security, user privacy protection, AI
   recommendation transparency, and false advertising control will be
   gradually implemented, guiding the industry from野蛮 growth towards
   standardized, high-quality development.
4. Rapid Diffusion to Mass Users: AI Shopping at a Critical Stage, Pain Points
   Remain Real
   In an interview with NBD reporters, Chen Yixin stated that current user penetration and
   usage of AI assistants are at a critical stage of “rapid diffusion from early adopters to the
   mass market.”
   “During the Spring Festival, products like Qianwen and Doubao successfully attracted
   massive user participation through various marketing手段. While this growth doesn’t
   directly equate to the use of shopping functions, it significantly lowers the barrier for
   users to接触 and use AI, completing初步 user education and habit cultivation,” Chen
   Yixin believes. When these users remain within the ecosystem, guiding them to more
   practical scenarios like shopping will result in a shorter conversion path.
   Since 2025, AI has accelerated its渗透 from back-end industrial applications to C-end e-
   commerce scenarios. Major platforms have strong impetus for the普及 and
   implementation of AI shopping.
   A former CTO of an e-commerce platform told NBD reporters that the Qianwen App is
   already抢占新一代 traffic gateway. Food delivery services can achieve a closed loop
   within the app, and more living service scenarios related to daily necessities will be

Integrated subsequently. He believes the landscape of the e-commerce industry will
gradually change, but accuracy and trust are crucial. “The content recommended by the
AI assistant must be what users truly need, and the price-quality ratio must also be
recognized by users.”

However, he also admitted that this process won’t happen overnight. Looking back at the
mobile internet era, various applications also went through long cycles from诞生 to
maturity, requiring not only the gradual improvement of infrastructure but also the
continuous cultivation and transformation of user mindsets.

Liu Yaping also pointed out real pain points in the actual implementation process.
First, the high concurrency and strong real-time nature of live streaming e-commerce
place extremely high demands on AI system stability. Model generalization capability in
complex scenarios is a challenge. Second, digital humans are still continuously being
optimized regarding emotional connection and improvisational interaction. Currently, they
are more suitable for standardized product explanation scenarios, complementing rather
than replacing human anchors. Furthermore, in terms of operational models, shifting from
relying on manual experience to data-driven decision-making requires a fundamental
transformation in team thinking and working methods.

Tian Feng, Dean of the Fast and Slow Thinking Research Institute, told NBD reporters
that the普及 of AI assistant shopping functions will take three to five years. “It will start
with young people and then permeate all age groups in society.”
Tian Feng believes that within one to three years, AI assistants will not replace traditional
e-commerce but will become core辅助 tools. E-commerce users will still rely on the
“Taobao/JD.com App main gateway,” but some users will first use AI assistants for辅助
decisions like “comparing products” during their购物 decision-making process. In the long
term (five years out), AI assistants will become the mainstream shopping gateway, “but
two conditions must be met: first, AI recommendation accuracy > 85%; second, seamless
integration of payment and after-sales service, such as ‘AI-generated return plans’,” Tian
Feng said.

3. Driving the Restructuring of the “E-commerce Industry” and “Consumption
   Logic”: AI Will No Longer Be an Independent Function
   Undoubtedly, AI will drive the restructuring of the e-commerce industry.
   “AI will no longer be an independent function but will be integrated into every环节:
   search, recommendation, customer service, marketing, design, logistics, etc. In the
   future, it will be difficult to distinguish whether a function is an ‘AI function’ because AI will
   be无处不在,” Chen Yixin believes. AI化 will become the infrastructure and core capability
   of shopping platforms, but its manifestations will diversify. The user’s shopping
   experience will deeply evolve from “people finding goods” to “goods finding people” or
   even “scenarios finding people.” Combined with multi-modal interaction (voice, gesture,
   image) and AR/VR technology, the shopping process will become more immersive,
   personalized, and convenient.
   Chen Yixin also stated that on the merchant side, AI tools will significantly lower the
   threshold and cost of operations, customer service, and content creation, enabling “千人
   千面” precision marketing and efficient conversion.
   (Image Source: NBD Media Library)
   Liu Yaping believes that AI shopping is not just the application of technological tools but a
   reconstruction of consumption logic. In his view, the core value AI brings to live streaming
   e-commerce and user consumption is reflected in multiple layers.
   AI can combine multi-dimensional factors like consumer preferences, shopping cycles,
   and lifestyle habits to achieve personalized recommendations, giving consumers a more
   precise shopping experience. Simultaneously, in links such as product selection, content

creation, and compliance review, AI greatly improves efficiency and reduces labor costs,
allowing teams to focus more on value creation. Technologies like digital human live
streaming and intelligent customer service make shopping more convenient and efficient.
AI real-time data analysis helps optimize live streaming strategies, enhancing user
satisfaction and conversion rates.

“AI is not a revolution of tools but an upgrade of cognitive paradigms. It is a lever amplifying human capabilities.”

The core of future competition will be AI literacy; those
who skillfully use AI will create greater value,” Liu Yaping said.
Furthermore, Liu Yaping pointed out that AI is promoting the evolution of live streaming
e-commerce from traffic conversion to value creation, forming a new pattern of
“technology driving efficiency, experience defining value.” For MCN agencies, the
opportunity lies in improving operational efficiency through standardized AI solutions, and
the change lies in the need to shift from人工经验-driven to data-driven decision-making.
At the same time, the monetization methods of e-commerce platforms will also undergo
significant changes.
Tian Feng believes that CPS will become the mainstream model, but it may be叠加 with
other innovative models.
He explains that the core value of AI shopping is improving conversion rates. Merchants
are willing to pay for “actual transactions,” not inflated exposure. “CPS deeply binds with
merchant interests; merchants are willing to pay for high-conversion traffic. This is the
underlying business logic and won’t change because of technology,” Tian Feng said. He
believes that since a large number of merchants generally pursue performance
advertising and order conversion rates, in the AI e-commerce shopping scenario, the
CPS model’s contribution to GMV will gradually surpass竞价排名 and member
subscriptions.

GEO in Ecommerce in China

Moreover, under this trend, GEO is expected to become the core方向 for subsequent
marketing transformation for brands. It will not only profoundly变革 brands’ online
customer acquisition logic and traffic operation models but also open up new growth
space for e-commerce service providers and cross-border e-commerce enterprises with
technical integration capabilities.
Related research reports from Guosen Securities mentioned that the monthly active
users of large models in China are currently close to 500 million, exceeding 1.27 billion
globally. Users are gradually forming the habit of searching for information on large
model platforms. Therefore, GEO, a new digital marketing strategy, has gained market
attention. Unlike traditional SEO, which optimizes keywords and rankings, GEO focuses
on whether brand content can be understood and recommended by AI large models.
Data from iResearch shows that in the second quarter of 2025, the GEO market size
grew by 215% year-on-year, and over 78% of corporate decision-makers listed AI search
optimization as a priority for digital transformation.
Furthermore, according to data from the China Academy of Information and
Communications Technology’s “2025 Generative Engine Optimization Industry White
Paper,” after GEO commercialization, the enterprise customer acquisition conversion rate
in AI recommendation scenarios increased by 2.8 times compared to traditional
searches, and the user decision-making cycle shortened by 40%. It is expected that GEO
will become a key focus for subsequent marketing transformation for brand merchants.
Reporter’s Notes: The Year of AI Shopping Popularization – A Business Revolution
Starting with Traffic and Ending with Infrastructure
The 2026 Spring Festival may become a special historical node: it marks AI shopping
officially crossing the chasm between “early adopters” and the “mass market.”

AI evolution is stirring in the depths of business logic. The rise of GEO is one of the most
noteworthy signals.

When China’s large model MAUs approach 500 million, and when third-party research
firm Gartner predicts traditional search engine traffic will decline by 25% in 2026, brands
have to face a fundamental question: If consumers no longer enter keywords through a
search box but instead ask AI to recommend products through conversation, where will
brand “visibility” come from?
The answer is rewriting the underlying logic of marketing.
From SEO to GEO, from “optimizing keyword rankings” to “making AI understand and
trust your brand,” this is a transfer of power over traffic distribution. In the future, brands
may all need to answer the same question: Can their content be understood and
recommended by Qianwen, Doubao, and Kimi?
And the ultimate指向 of all this is AI evolving from a “function” into shopping
“infrastructure.”
Perhaps the day we stop talking about “AI shopping” will be the moment this
transformation is truly complete. Just as we don’t say “electric life” or “internet society”
today because they are already everywhere.
Back to the present, as subsidies recede, most users remain in the information search
stage and have not yet formed the habit of completing purchases within AI assistants.
This also reminds us that the future has only just begun.

Read also

Top E-Commerce Trends in China