Walmart Opens Shopping Access to AI Agents—Why Is Amazon Holding Back?

Walmart is doing something traditional retail giants have never done before: letting AI agents shop the supermarket for you. According to RSS sources, the retail giant will next month allow users to connect their own AI agents to Walmart's website, complete with controls such as purchase approval and spending limits. This move stands in sharp contrast to Amazon's cautious stance and pushes "AI agent shopping" from concept into a real battlefield.

Walmart's AI Agent Shopping: Open but with Boundaries

Walmart's approach does not let AI agents take full control of accounts. Users need to actively connect their own AI agents to the Walmart platform, and the system provides purchase approval and review processes along with support for spending caps. In other words, AI can handle the tedious parts of selecting and placing orders, but the final control over the wallet remains with the user.

This design philosophy is noteworthy: it unlocks the automation value of AI agents while reducing the risk of losing control through manual approval and budget constraints. For retailers, opening the AI agent entry point means transactions may be completed without users directly visiting the website, which requires the platform to do substantial adaptation work on identity verification, order confirmation, and after-sales processes.

AI Agent refers to software programs capable of perceiving their environment, making autonomous decisions, and executing tasks. Unlike traditional chatbots, they can call tools, access external APIs, and complete multi-step operations. In shopping scenarios, AI agents can browse products, compare prices, add items to cart, and even complete orders on behalf of users, who only need to set goals and constraints. The core value of this "agentic shopping" lies in automating repetitive labor, but it also introduces authorization boundary issues—to what extent can agents make consumption decisions on behalf of users, and how to ensure their behavior aligns with users' true intentions. Walmart's requirement for users to actively connect and set approvals and limits is essentially drawing a safety zone between technical capability and user control.

Divergence from Amazon's Approach

According to reports, Walmart's embrace of AI agents forms a "sharp contrast" with Amazon. As a dominant player in e-commerce infrastructure, Amazon has maintained a conservative attitude toward third-party AI agents accessing its shopping processes. Behind this divergence is a collision of two business logics.

Walmart's choice to open up may stem from a challenger posture: by lowering the barrier for AI agent access, it attracts users accustomed to intelligent assistants and turns AI agents into a new traffic entry point. Amazon, meanwhile, prefers to internalize AI shopping capabilities into its own assistant system, avoiding third-party agents weakening its influence over users' shopping decisions.

Open APIs and agent access on e-commerce platforms are key background for understanding the divergence between Walmart and Amazon. Open APIs allow third-party developers or AI agents to programmatically access platform product catalogs, inventory, prices, and ordering interfaces, thereby embedding the shopping process into external assistants. Amazon has long been cautious about third-party agent access, partly because it has already built its own assistant ecosystem such as Alexa and prefers to keep AI shopping capabilities closed-loop to maintain control over user data and transaction entry points. Walmart lacks a comparable in-house assistant system, so opening agent access can be seen as a "leverage" strategy: using users' existing AI tools to expand reach while retaining the platform's final review authority over transaction processes through approval and limit mechanisms. This route difference reflects the different priorities of challengers and incumbents regarding ecosystem control.

The Real Problems AI Agent Shopping Must Solve

Letting AI agents spend money on people's behalf involves trust barriers beyond technology. Spending limits and approval mechanisms address fund security, but product quality judgment, return and exchange responsibility, and price fluctuation response still require platforms and agent developers to jointly establish rules.

From the information Walmart has released, its focus is on a control framework of "approvable, limitable," suggesting the platform views AI agents as automated buyers that need constraints rather than fully trusted independent decision-makers. This pragmatic attitude may become an industry reference template—first let agents run, then gradually open up permissions.

Machine-readable product information is an important prerequisite for AI agents to shop efficiently. Traditional e-commerce pages are designed for human vision, containing large amounts of images, promotional copy, and vague descriptions, whereas AI agents need structured data—such as standardized product attributes, real-time prices, inventory status, delivery times, and return policies—to make reliable comparisons and decisions. If platforms do not provide machine-readable interfaces or structured data, agents can only rely on web scraping, which is prone to failure due to page changes or anti-scraping mechanisms. Therefore, Walmart's opening of agent access is not just a permissions issue but also involves infrastructure transformation at the product data layer, and brands need to adjust how they organize product information to make it machine-friendly.

Impact on Retail and the AI Ecosystem

If Walmart's pilot goes smoothly, AI agents could become a new "customer type" for e-commerce platforms. Brands need to think about how to make product information machine-readable and comparable; agent developers need to handle cross-platform identity and payment integration; and platforms need to find a balance between openness and control.

Whether Walmart's move works depends on the actual experience and user acceptance after launch next month. But at least it sends a clear signal: AI agents are no longer just chat tools—they are becoming real transaction participants.

(This article is based on public information from RSS sources; specific feature details are subject to Walmart's official announcements.)

One-line Summary

Walmart will open its shopping site to AI agents next month with approvals and limits, contrasting with Amazon's conservative stance.

Paragraph Summary

Walmart plans to allow users to connect their own AI agents to its website for shopping next month, with controls such as purchase approval and spending limits, letting AI handle selection and ordering while the user retains final control of the wallet. This openness contrasts sharply with Amazon's conservative attitude toward third-party AI agents, reflecting different business logics: challengers use open entry points to gain traffic, while incumbents prefer closed-loop control. The article notes that for AI agent shopping to truly take off, beyond fund security, it must solve trust and rule issues such as product quality judgment, return responsibility, and price fluctuation response, and platforms need to balance openness and control. If the pilot succeeds, AI agents could become a new customer type for e-commerce platforms, driving machine-readable brand information and cross-platform identity and payment integration.