The early excitement surrounding agentic commerce — the idea that artificial intelligence (AI) agents can search for products, compare options and complete purchases on behalf of buyers — is encountering a reality check as major technology companies refine their strategies.
Recent moves from OpenAI and Amazon suggest the first generation of AI-driven purchasing may evolve more slowly than early demonstrations suggested.
Instead of replacing ecommerce platforms and procurement systems, companies are increasingly positioning AI agents as tools that assist buyers with product discovery and purchasing workflows. Meanwhile, transactions remain inside merchant and enterprise systems.
For B2B ecommerce, orders often involve negotiated pricing, credit terms, ERP integrations and complex logistics. As a result, the shift underscores how difficult fully autonomous purchasing will be to implement at scale.
And for B2B ecommerce platforms, the approach aligns more closely with how transactions already occur. Those platforms typically integrate with enterprise resource planning systems, contract pricing databases and inventory management tools.
Allowing an external AI interface to execute transactions directly could bypass financial and operational controls embedded in distributor and manufacturer systems.
How OpenAI has invested in agentic commerce
OpenAI was among the most visible proponents of agentic commerce in 2025, introducing shopping capabilities within ChatGPT and a feature known as Instant Checkout, which allowed some users to complete purchases directly within a chat session.
The company also collaborated with payment provider Stripe to introduce the Agentic Commerce Protocol. They designed it to allow AI agents to securely interact with merchant systems and initiate transactions.
More recently, however, OpenAI has begun steering developers toward routing purchases through merchant-controlled checkout systems rather than completing transactions inside ChatGPT.
Under that approach, ChatGPT functions primarily as a discovery and recommendation layer that helps buyers locate products and connect with merchants. Payments, taxes, order management and customer service remain within the merchant’s existing commerce infrastructure.
Amazon tests the waters with AI agents
Amazon has also been experimenting with agent-driven purchasing, though the company has taken a cautious approach to expanding its capability.
The retailer introduced a feature known as Buy for Me within the Amazon shopping app that allows its AI systems to place purchases from external brand websites when products are not available directly through Amazon.
The system can search external retail sites, complete checkout details and confirm the order with the customer.
The feature remains in beta and is available only to a limited group of users. In addition, the external retailer — not Amazon — is responsible for shipping, returns and customer service.
Amazon has also expanded AI tools inside its own marketplace ecosystem through Rufus, a generative AI shopping assistant designed to answer product questions and guide purchase decisions.
Together, those initiatives highlight Amazon’s interest in AI-driven shopping while illustrating the operational challenges associated with allowing AI systems to transact across multiple retail platforms.
Those challenges are magnified in B2B commerce.
Agentic commerce in a B2B ecommerce setting
Distributor and manufacturer transactions frequently include negotiated pricing, customer-specific catalogs, freight calculations, tax exemptions, credit approvals and order workflows tied directly to ERP systems.
Orders can involve dozens or hundreds of line items and represent significant dollar values, increasing the risks associated with automated purchasing decisions.
Industry research suggests adoption remains in its initial stages. Consulting firm Deloitte reports that fewer than one-quarter of B2B suppliers currently use agentic AI technologies, even as many companies plan to increase artificial intelligence investments during the next two years.
Many organizations are also still developing governance frameworks that determine how much authority AI systems should have over purchasing and financial transactions.
For distributors and manufacturers, the near-term opportunity in agentic commerce may lie in AI-assisted workflows rather than fully autonomous purchasing.
Examples include contract-aware product search, automated reorder recommendations, quote-to-order assistance, and procurement tools embedded within ecommerce portals.
These capabilities allow companies to use AI to streamline B2B purchasing while keeping the final transaction within existing ecommerce and enterprise systems.
The recent shifts by Amazon and OpenAI suggest the initial narrative around agentic commerce is entering a more practical phase.
Instead of replacing traditional ecommerce platforms, AI agents are more likely to become an embedded layer of automation within the digital infrastructure that already supports B2B buying and selling.
Sign up
Sign up for a complimentary subscription to Digital Commerce 360 B2B News. It covers technology and business trends in the growing B2B ecommerce industry. Contact Mark Brohan, senior vice president of B2B and Market Research, at mark@digitalcommerce360.com. Follow him on Twitter @markbrohan. Follow us on LinkedIn, X (formerly Twitter), Facebook and YouTube.