By Wang Hong and Ricky Tan

In October 2025, Walmart announced a landmark partnership with OpenAI to bring shopping directly into ChatGPT. Through a new Instant Checkout feature, Walmart customers can browse, select, and complete purchases entirely within the ChatGPT interface, with Walmart managing fulfilment. The collaboration reframes the shopping journey—from “search, redirect, and order” to “conversation, confirmation, and transaction”—marking the transition of AI agents from information gateways to transaction enablers.

Walmart described the initiative as an “AI-first” retail experience, introducing the concept of “agentic commerce”: a future in which intelligent agents act on behalf of consumers to complete purchases seamlessly. Public information indicates a phased rollout across Walmart and Sam’s Club, with payment support from partner providers.

This development mirrors a broader global movement. In China, companies are laying the foundations for the intelligent-agent era within a distinct ecosystem and regulatory environment. They are building infrastructure that connects AI systems to real-world commerce through verifiable, engineering-driven approaches—integrating core capabilities such as content understanding and cross-platform product knowledge bases into standardised interfaces linking front-end agents with back-end transaction systems.

As AI permeates industry, its impact on online retail is particularly visible. AI is evolving from an information assistant into a full transaction executor, shifting e-commerce logic from passively responding to searches to proactively understanding and fulfilling user intent. This represents a systemic upgrade for e-commerce—from increasing conversion rates to ensuring end-to-end transaction completion and strengthening consumer trust.

A vivid illustration of this transformation can be seen in Zhidemai Technology, founded by CEIBS DBA graduate Sui Guodong. Through observation, interviews, public information and analysis of company practices, we developed a teaching case capturing these trends. Looking at both the Walmart–OpenAI model and Zhidemai’s experience, one conclusion is clear: the future of ecommerce competition in China will depend not only on products and services, but on the ability to build and govern intelligent-agent ecosystems.

Walmart × OpenAI: From Information Search to Transaction Execution

Walmart CEO Doug McMillon framed the partnership with OpenAI as a pivotal moment in online shopping, saying: “For many years now, ecommerce shopping experiences have consisted of a search bar and a long list of item responses. That is about to change.” His words capture the essence of this shift, towards conversational, intent-driven, AI-executed commerce.

1. Instant Checkout: End-to-End Transactions via Ecosystem Collaboration

According to The Wall Street Journal, ChatGPT users in the US will soon be able to purchase nearly all Walmart products—except fresh groceries—directly in chat. Walmart+ benefits, including free shipping, will also be available.

The transformative power of this lies in closing the gap between decision and execution within a single interface. Previously, AI could make recommendations, but users still had to navigate to an ecommerce site to complete the purchase. Now, a query like “What do I need for a weekend barbecue?” can result in an AI-generated shopping list and an instantly placed order after user confirmation.

This evolution—from AI assistant to transaction executor—underscores the importance of ecosystem partnerships. As Zhidemai CTO Wang Yunfeng notes, even if each step of a long multi-step process has a 95% success rate, the overall success probability drops to about 36% across 20 steps. AI alone cannot guarantee consistent performance in long-chain transactional tasks. Deterministic capabilities from ecosystem partners are essential.

Thus, the Walmart x OpenAI collaboration does more than provide product access. It equips AI with reliable transactional capacity, redefining the ecommerce experience.

2. Walmart’s Strategic Intent: Becoming the AI-Era Traffic Gateway

As consumers turn to AI for shopping decisions, Walmart aims to become the preferred fulfilment partner embedded within AI interfaces. In the mobile internet era, traffic gateways were dominated by search engines and super apps. In the AI era, platforms that best understand intent and execute tasks will become the next-generation gateways.

By integrating deeply with leading AI platforms, Walmart embeds its capabilities within the emerging AI operating system, ensuring that no matter where consumers initiate a request, Walmart remains the go-to fulfilment provider.

3. A Strategic Concession: Data Sharing with AI Platforms

To seize this opportunity, Walmart made an unprecedented move: allowing transactions to occur entirely within ChatGPT while also granting OpenAI access to its purchase data.

For a retailer of Walmart’s scale, purchase data has traditionally been a core asset. Sharing it signals a new balance of power in the AI-driven economy—retailers may need to trade some data sovereignty for AI ecosystem access. This also hints at deeper shifts in industry value distribution.

The Rise of AI Agents and Their Retail Impact

Walmart’s partnership with OpenAI reflects how AI agents are reshaping retail worldwide. Many analysts consider 2025 the “first year of AI agents.” Unlike earlier large language models (LLMs), AI agents can interpret complex goals, plan independently, and use tools to complete cross-application tasks. This dramatically affects consumer behaviour and the broader retail ecosystem.

1. The Outlook of Agentic Commerce: Beyond Retail

Agentic commerce will transform business far beyond consumer retail. In B2B operations, AI agents already manage supply chains, perform financial tasks, and deliver problem-solving customer service. Developers use AI to write and debug code autonomously; research agents conduct market research and data analysis.

E-commerce is the most visible frontier because its transactional loops are clear and user engagement is high. Yet it represents only the beginning of what we refer to as the AI Agent Economy.

2. Reshaping Consumer Behaviour: Decision Outsourcing and Frictionless Shopping

i) AI as a Decision-Making Gateway

A recent study by the National Bureau of Economic Research in the US, How People Use ChatGPT, found that as of July 2025, around 10% of adults around the world had used ChatGPT, with over 70% of usage being non-work related. Practical guidance, information seeking, and writing accounted for 80% of interactions.

Consumers are increasingly outsourcing decisions to AI. In shopping, this means relying on agents to navigate the complexity of search and comparison. Zhidemai’s observations support this: in 60% of consumer–AI interactions, users were essentially asking, “What should I buy?”

ii) Frictionless Shopping and the Intention Economy

AI agents are enabling the “Intention Economy,” in which consumers express intent and agents fulfil it automatically—sometimes invisibly. Mr. Sui describes a speculative scenario in which a smart fridge told to “replenish eggs,” for example, triggers cross-platform ordering via Model Context Protocol (MCP) interfaces and confirms the purchase within minutes.

iii) AI Redefines Consumer Loyalty

As AI agents become the primary interface for shopping, consumer trust and loyalty shift from retailers to the agents themselves. Companies must prioritise agent trustworthiness—security, transparency, and user control will be decisive.

Ecosystem Disruption: Disintermediation and the Infrastructure Race

i) Traditional intermediaries—including ecommerce platforms and search engines—face disruption as agents connect demand directly with supply chains. Platforms must evolve into agent-driven ecosystems or risk becoming mere fulfilment backends.

ii) The Battle Over AI Standards

To execute transactions, agents need secure interfaces linking them to product catalogues, inventory, payment, and logistics systems. Industry standards such as Anthropic’s MCP and Google’s Agent-to-Agent (A2A) architecture are therefore emerging to fill the gap.

This competition resembles those between PC operating system and mobile app ecosystems in the past. Whoever sets the standards will shape how agents interact with the commercial world.

iii) The Return of Centralised Data Power

AI agents will accumulate unprecedented cross-context consumer data—from shopping habits to daily routines. Regulators face major challenges in balancing innovation with privacy protection.

China’s Strategy: Infrastructure Beyond Closed Ecosystems

As AI reshapes global retail, China is charting a distinctive path. Companies like Zhidemai are adapting global trends to China’s market dynamics, platform structures, and regulatory environment. The company provides an instructive example as to how Chinese enterprises are harnessing developments in AI in a way that is distinct from their global competitors.

1. Zhidemai’s Strategy: Engineering Infrastructure for the Agent Era

Zhidemai embraced AI early, making AI-generated content (AIGC) a strategic priority in 2023 and launching a comprehensive AI strategy in 2024. After two years of exploration, its vision now extends beyond building a consumer-facing super app. Instead, it aims to create a high-quality, collaborative AI ecosystem and foundational infrastructure for the agent era, centred around its “Haina” MCP Server.

i) From Content Insight to Ecosystem Integration

Zhidemai’s strategy leverages two core assets:

• a cross-platform repository of premium products and content, built over a decade;
• its proprietary AI Understand Content (AIUC) engine.

The Haina MCP Server exposes these capabilities externally, connecting front-end agents with back-end consumer data and transaction systems—effectively acting as a “USB interface” for AI commerce.

ii) A Bold Vision for an Open Ecosystem

Application Programming Interfaces (APIs) are sets of rules and protocols that allow different software applications to communicate with each other and exchange data. Since launching in May 2025, Haina’s monthly API calls have surged sixfold to nearly 30 million. Partnerships span leading players such as Huawei’ Xiaoyi, Tencent’s Yuanbao, and Moonshot’s Kimi.

Commercialisation follows a classic ecosystem playbook:

1. free API access to accelerate adoption;
2. a commission-based Sales Alliance;
3. an Advertising Alliance to monetise traffic.

2. Contrasting Models: Vertical Integration vs. Horizontal Collaboration

The Walmart x OpenAI model and Zhidemai’s approaches represent two distinct paradigms for building business ecosystems in the AI agent economy:

i) Walmart x OpenAI: A Vertically Integrated Powerhouse

A closed-loop partnership combining OpenAI’s front-end intelligence with Walmart’s fulfilment infrastructure. The model offers superior experience but is limited to Walmart’s ecosystem.

ii) Zhidemai: Horizontally Open Platform

Zhidemai connects multiple agent front ends with multiple ecommerce back ends. Its goal is to be a neutral middle layer bridging agents and commerce infrastructure. The company is partnering with Chinese cloud and marketing services provider Weimob to develop a Chinese agent business protocol, addressing identity, payments, and logistics interoperability. This model’s strength is openness and scale potential; its challenge is maintaining independence while coordinating diverse stakeholders.

3. Navigating China’s “Walled Gardens”

In China, agentic ecommerce must adopt a localised strategy to navigate the market’s unique internet ecosystem, regulatory framework, and consumer behaviour.

i) Ecosystem Barriers and the Value of Data

China’s major platforms operate isolated ecosystems. This creates challenges but also elevates the strategic value of neutral players like Zhidemai. Cross-platform product data, rare within closed ecosystems, becomes a crucial asset.

ii) Innovation Under Constraints: The “Virtual Machine + Agent” Approach

To operate across partially closed interfaces, companies combine workflow orchestration, rule-based routing, and human-machine collaboration. Priority goes to securing critical links such as identity, payments, and fulfilment, while high-risk processes undergo secondary verification.

iii) Trust and Compliance

Chinese regulation, including the Personal Information Protection Law (PIPL), demands strict data governance. Zhidemai’s next-generation agent, Zhang Dama, uses “fully traceable” mechanisms allowing users to inspect every AI action, which is key for earning trust and meeting compliance requirements.

iv) Super Apps as Gatekeepers

WeChat, Douyin, and Taobao are integrating AI and agent capabilities, creating stiff competition. Zhidemai’s approach of partnering with model providers and device manufacturers rather than competing directly for traffic reflects a strategic “co-opetition” mindset.

IV. Strategic Takeaways for Managers

The Walmart x OpenAI partnership marks the beginning of AI-powered ecommerce at scale. Companies worldwide are taking different paths: primarily vertically integrated models in the US and horizontally open platforms in China. Managers navigating this shift should focus on four priorities:

1. Clarify your ecosystem position.

Decide whether to operate as a front-end agent gateway, back-end fulfilment provider, or infrastructure builder. Most retailers should aim to be “preferred partners” for agent platforms.

2. Leverage data for competitive advantage.

High-quality, proprietary data—such as Zhidemai’s cross-platform product base and AIUC engine—will differentiate winners in an increasingly homogenised retail market.

3. Foster open collaboration.

No company can build the agent ecosystem alone. Collaboration across platforms and contribution to industry standards are essential, particularly in markets with “walled gardens” like China.

4. Balance agile innovation with steady governance.

As AI evolves rapidly, companies must innovate while maintaining strategic flexibility and financial discipline. Zhidemai’s balancing act between consumer applications and B2B ecosystem-building exemplifies this challenge.

Whether we like it or not, agentic ecommerce is here, and it is redefining not only the shopping experience but also the underlying power structures of global business. To thrive, companies must embrace this shift, understand its logic, and position themselves strategically within the emerging ecosystem. With data-driven capabilities, collaborative standards, and proactive governance, China’s ecommerce players and global retailers alike can forge their own paths in the rapidly emerging AI agent economy.

Wang Hong is President and Professor of Management at CEIBS; Ricky Tan is Professor of Decision Sciences and Management Information Systems at CEIBS.