Tech · 22 October 2026 · 4 min read
While Amazon Blocks AI Agents, Walmart Welcomes Automated Shoppers via MCP
In brief: While major e-commerce platforms like Amazon actively block autonomous AI shopping agents, Walmart is taking the opposite route. The US retail giant has launched Agent Gateway, an integration layer based on the open Model Context Protocol (MCP) that enables vetted AI agents to search inventory and complete purchases under strict spending and privacy guardrails.
by Team Mocchi's
In the emerging landscape of agentic commerce — where autonomous software handles product discovery, decision-making, and financial checkout on behalf of consumers — retail powerhouses are picking opposing strategies. While Amazon has escalated technical countermeasures to block third-party shopping agents, Walmart has officially opened its digital storefront to AI-driven transactions.
Speaking at the Masters of Scale Summit in San Francisco, Walmart US President and CEO John Furner introduced a suite of tools built to accommodate external agents such as OpenAI's ChatGPT, Google's Gemini, and Meta's Muse, extending compatibility across both Walmart and Sam's Club.
An Infrastructure Built on MCP Standards
At the technical core of Walmart's initiative is the Agent Gateway, designed as a dedicated server running the Model Context Protocol (MCP). Originally created by Anthropic, MCP provides an open standard for large language models to interact cleanly with external APIs, databases, and enterprise tooling.
As reported by TechCrunch, rather than forcing autonomous agents to rely on brittle web-scraping routines and headless browsers, Agent Gateway establishes a structured, documented interface. Approved agents can directly query current stock, compare pricing across local fulfillment centers, and submit purchase requests through machine-readable channels.
Privacy Abstraction and Spend Controls
Delegating purchasing power to artificial intelligence introduces critical privacy and financial safety challenges. To mitigate these risks, Walmart built an abstraction layer that isolates personal customer records from external AI providers.
When a shopper instructs an agent to restock staple groceries, Walmart's gateway resolves the request against the user's past purchase habits internally. The AI receives only the necessary product match, shielding full order histories and credit card details from the foundational model provider.
Furthermore, as highlighted by The Verge, the system introduces dedicated Agent Profiles. Shoppers can configure granular spending boundaries, such as setting pre-approved thresholds up to $100 or requiring manual biometric authorization for every transaction. Walmart's terms strictly bar automated agents from modifying saved payment methods or performing actions that deviate from explicit user intent.
Walled Gardens Versus Open Standards
The divergence between Amazon and Walmart reflects a fundamental debate over the future of online platforms. Amazon and several online travel agencies treat autonomous agents as hostile crawlers that threaten retail media networks and dilute direct customer engagement.
Walmart, by contrast, is attempting to capture transaction volume at the point of intent generation. By formalizing technical specifications and sandboxing agent actions, the retailer aims to transform programmatic shopping from an unmanageable security challenge into a verified sales channel.
Mocchi's take
Walmart's deployment of MCP illustrates how quickly open agent protocols are migrating from developer environments into mission-critical transactional platforms. For European businesses and digital retailers, this shift proves that optimizing exclusively for human web traffic and traditional search engines will soon be insufficient. Preparing API architectures to interact securely with autonomous software agents is becoming an essential roadmap requirement, demanding rigorous data boundaries to prevent third-party platform lock-in.