Google sets open protocol for AI shopping

Google has unveiled a Universal Commerce Protocol designed to allow artificial intelligence agents to search, recommend and complete purchases across multiple retail platforms, marking a strategic push to standardise how online shopping works in an era of increasingly autonomous digital assistants. The announcement was made at the 2026 National Retail Federation conference, where technology providers and merchants gathered to outline priorities for the year ahead.

The protocol is presented as an open standard intended to connect product discovery, pricing, inventory checks, payments and fulfilment into a single interoperable framework. Google said the initiative has been developed with a group of early partners that includes e-commerce platforms, payments networks and logistics providers, with the stated aim of reducing friction created by fragmented systems that dominate global online retail.

Google’s move to standardise agent-led commerce was positioned as a response to rapid advances in generative AI and autonomous agents, which are increasingly being deployed to act on behalf of consumers. These agents can already compare prices, summarise reviews and manage shopping lists, but they often fail at the transaction stage because each retailer operates its own technical and contractual setup. The Universal Commerce Protocol is designed to give AI agents a common language to complete purchases securely without bespoke integrations for every merchant.

According to Google executives presenting at the conference, the protocol defines shared rules for identity, consent, product data, checkout flows and post-purchase services such as returns and customer support. Merchants adopting the standard would expose a set of APIs that authorised agents can access, while retaining control over pricing, availability and brand presentation. Consumers, in turn, would be able to instruct AI assistants to buy items within defined preferences and budgets, rather than navigating multiple apps or websites.

The initiative reflects a broader shift in retail technology toward “agentic” commerce, where software acts with a degree of independence once parameters are set. Industry analysts note that while recommendation engines and chat-based shopping tools have become commonplace, fully automated purchasing has been constrained by security concerns and the lack of agreed standards. Google argues that an open protocol can address both issues by embedding verification, audit trails and revocation mechanisms from the outset.

Payments integration is a central element of the design. By working with established payments networks, the protocol seeks to ensure that AI-initiated transactions comply with existing financial regulations and fraud controls. Tokenisation and user-approved credentials are intended to prevent agents from accessing sensitive card details directly, a point Google highlighted as critical to building trust among both consumers and regulators.

For retailers, the promise lies in greater reach and personalisation. An AI agent using the protocol could surface a merchant’s products to consumers who might never visit its website, while tailoring offers based on explicit user preferences rather than opaque behavioural tracking. Smaller retailers, in particular, could benefit from reduced technical barriers to participating in AI-driven marketplaces, though adoption will depend on costs and the perceived balance of power between platforms and sellers.

The announcement also underscores intensifying competition among technology groups to shape the future of digital commerce. Several rivals are developing proprietary shopping agents tightly coupled to their ecosystems. Google’s decision to frame its approach as open and collaborative appears aimed at attracting broad industry support and avoiding accusations of gatekeeping. Whether the protocol remains genuinely neutral will be closely watched by regulators already scrutinising the influence of large technology companies in advertising and retail.

Consumer advocates have raised questions about transparency and accountability when purchases are delegated to machines. Google says the protocol requires explicit user consent for each category of action and allows consumers to review, approve or cancel transactions initiated by agents. Detailed logs are intended to make it clear why an agent chose a particular product or seller, addressing concerns about hidden bias or undisclosed commercial incentives.