How retailers can decentralize AI shopping without losing customer context

AI shopping is changing where retail discovery, customer context, and transactions happen. The executive decision is how to divide those responsibilities across external AI interfaces and retailer systems. Retailers can decentralize selected parts of the customer journey while keeping accounts, loyalty, transaction processing, fulfilment, and support connected to their own commerce architecture.

Retailers use websites, apps, accounts, and loyalty programmes to connect browsing and transactions with customer history. Emerging integrations can divide those functions among several systems. An AI service can handle discovery or present checkout while retailer systems retain selected customer context, loyalty connections, and transaction responsibilities.

The choice is shifting toward selective decentralization

Google and OpenAI are building commerce arrangements that span different systems. This creates selective decentralization: moving selected parts of the commerce journey to external interfaces while keeping chosen retailer systems connected. For CEOs and CTOs, the question is which responsibilities should move and which systems must remain connected.

“Control” therefore breaks into specific decisions. One company can determine what shoppers see. Another can hold account and loyalty information, while retailer systems calculate tax, process payment, fulfil the order, and support the customer. Merchant-of-record status means the party legally responsible for selling the goods in a transaction; interface ownership, behavioral-data access, and use of customer history are separate questions.

External discovery can provide access to demand while retailer systems continue to handle selected functions. Executives need to determine which party sees customer information, recognizes an account, applies loyalty benefits, processes the transaction, fulfils the order, and supports the customer afterward. Treating those responsibilities separately makes the architecture and commercial trade-offs easier to assess.

AI shopping divides commerce responsibilities

Wayfair illustrates the operational split between an external interface and retailer infrastructure. The arrangement separates legal responsibility from interface ownership: merchant-of-record status identifies where transaction responsibility sits, while an external service can provide the customer-facing interface. Executives assessing similar arrangements need to evaluate data access, account recognition, loyalty, support, and post-purchase interactions as distinct design decisions.

OpenAI is testing a related architecture through its Agentic Commerce Protocol. The broader design question is how payment providers, fulfilment logic, carts, loyalty recognition, and checkout design fit together when an external AI service participates in shopping. OpenAI has a commercial interest in making ChatGPT useful for product discovery and commerce, so executives should assess its architecture in that context.

Retailers are also developing AI experiences on their own properties. Ulta Beauty benefits commercially from reaching shoppers across external and owned interfaces. Walmart is pursuing a similar cross-system model, while Walmart and Google each benefit from deeper shopping activity in their respective systems. These incentives matter when executives evaluate how responsibilities and customer relationships are divided.

Customer context and loyalty can travel across interfaces

Customer recognition can remain useful when discovery happens elsewhere. Browsing behavior can indicate interest, baskets can show combinations under consideration, and completed transactions record purchases. Accounts and loyalty programmes can connect those interactions over time. The architecture question is whether an external AI experience can use relevant retailer-held context with the permissions and system connections that experience requires.

Retailer-held context can become an input to an experience delivered through an external AI interface. The service presenting discovery can influence which products appear and which information frames the shopper’s evaluation. Retailer-held account and loyalty data can preserve continuity across that boundary, while each implementation determines which information moves between companies and how it is used.

Portable AI commerce requires structured product data and connected systems

Selective decentralization creates an infrastructure requirement. Product information needs a machine-readable structure so external systems can interpret it, while price and availability data need regular updates. Deeper integrations can also require connections to accounts, loyalty, carts, payments, fulfilment, and checkout. AI shopping therefore reaches core commerce systems as well as customer acquisition.

Structured feeds provide one way to supply product information to external AI services. For CTOs, feed quality becomes part of the distribution architecture. Incomplete attributes or stale inventory can limit the product information available through an integration. The same discipline applies to the interfaces connecting external services with customer and transaction systems.

The deeper engineering task begins when an AI interaction needs customer state. Linking an external interaction to an account, applying loyalty benefits, recognizing an existing cart, and sending an order through checkout require system interfaces and explicit rules for data exchange. Retail technology leaders can act now by mapping each commerce responsibility to its system of record, defining what customer and transaction data may cross company boundaries, and testing whether product, account, loyalty, cart, and checkout interfaces can support that division.

Main highlights

Select which commerce functions to decentralize: Treat discovery, account recognition, loyalty, payments, fulfilment, and support as separate responsibilities. Leaders should decide which can move to external AI interfaces and which must remain connected to retailer systems.
Define responsibilities across AI shopping systems: Merchant-of-record status does not determine interface ownership or data access. CEOs and CTOs should establish who controls each customer, transaction, and post-purchase function.
Make customer context portable with clear boundaries: Account and loyalty data can preserve continuity when shopping begins in an external AI interface. Leaders should define what customer information can cross company boundaries and how external services may use it.
Prepare core commerce infrastructure for AI distribution: AI shopping depends on structured product data, current price and inventory information, and interfaces to accounts, loyalty, carts, checkout, and fulfilment. CTOs should map systems of record and test these connections before deeper integrations.