Agentic AI use cases are emerging across retail, impacting not only how companies operate an eCommerce engine but also how consumers interact with agents to automatically buy products or personalize browsing.

And the trend is moving at warp speed.

Take Walmart’s partnership with OpenAI and ChatGPT, which, since the partnership was first announced in late 2025, ChatGPT reportedly already accounts for 20% of Walmart’s referral traffic online.

Agentic AI is becoming a way of business for eCommerce retailers, but to keep up with the trend, retailers must develop a technology backbone that supports a team of agents and their long-term success.

What Agentic AI Looks Like Within eCommerce

First, agentic AI and multiple AI agents working across departments are not chatbots or surface-level automation; they are autonomous software entities that can reason, collaborate with other agents, and take action across systems. This means agents can update pricing and promotions online in real time, forecast inventory based on weather patterns or changes in shopper behaviors, and automatically reorder and purchase products for shoppers.

Yet, the essential insight here for retail executives is that AI agents do not operate independently. They function as coordinated systems of intelligence, a group of tens to hundreds of AI agents, depending on the operation.

This reality today changes the requirements needed for an underlying technology architecture.

Agentic AI adoption empowers multiple AI agents working together. Merchandising agents collaborating with inventory and fulfillment agents to determine not only what to promote, but where and when online. Another example: loyalty agents tailoring offers based on customer lifetime value, local assortment, and margin targets, and then coordinating execution with commerce and content agents.

This shift is not incremental efficiency. It represents a new operating model for retail. And it requires different foundations than monolithic platforms. The transition won’t happen overnight, but the architectural decisions made today determine whether organizations can scale with this transformation.

Agentic AI will not scale on monolithic platforms; retailers must adopt composable, API-driven architectures that allow coordinated AI agents to operate securely, intelligently, and autonomously across the business.

Keys Toward Underlying Architectures

Composable architecture, guided by MACH principles that emphasize microservices, API-first design, cloud-native infrastructure, and headless systems, is the foundation that agentic computing requires.

Composable platforms allow retail capabilities to be exposed as services that AI agents can consume, reason over, and coordinate. This includes:

• API-first microservices that allow agents to execute discrete retail functions such as pricing, promotions, inventory checks, and returns.

• Standardized open APIs that enable interoperability across internal teams, external partners, and third-party agent ecosystems.

• Governed orchestration that ensures agents operate within defined policies, security constraints, and observability frameworks.

• Modular design that allows new agents to be introduced or retired without replatforming the business.

Forthcoming research from the MACH Alliance reinforces this direction. According to the organization’s global annual research report, being released in February 2026, 75% of retail, commerce, and eCommerce leaders have widely or fully adopted composable and MACH principles. This compares to 67% in manufacturing and 62% in financial services, highlighting retail’s leadership in composable adoption.

The same research reveals a more nuanced picture of AI maturity, with 69% of retail leaders reporting that their companies are actively using AI as an integrated component of their technology stacks. This refers to embedded AI capabilities within platforms and workflows, not standalone tools such as ChatGPT or AI features provided indirectly through vendors.

At the same time, 22% of global leaders surveyed do not have AI integrated into their core technology stacks at all. These organizations rely on open applications like Copilot or ChatGPT, or encounter AI only through vendor partners that use it. This is AI by proxy, not AI as a strategic capability.

This distinction matters. Agentic computing cannot be achieved through surface-level exposure to AI. It requires intentional architectural design and deep integration into the retail operating model.

Building for What Comes Next

AI agents will transform retail operations and customer experiences, but only for organizations that invest in the right foundations. Monolithic platforms were designed for centralized control and predictable workflows. Agentic systems thrive on modularity, interoperability, and governed autonomy.

For retail executives, the mandate is clear. Composability is no longer optional. It is the infrastructure that allows agentic strategies to scale responsibly across channels, categories, and customer journeys.

The rise of agentic computing is a defining moment for retail. Those who align architectural discipline with strategic ambition will unlock its full potential. Those who do not risk building intelligent systems on foundations that cannot support them.