NEW YORK — The rapid expansion of AI-driven commerce is revealing significant vulnerabilities in traditional e-commerce checkout systems not designed for machine-speed interactions, according to industry experts.
Orders originating from AI-powered search on platforms like Shopify have increased 15-fold since January 2025, routing through protocols such as Google’s Universal Commerce Protocol (UCP) and OpenAI’s Agentic Commerce Protocol (ACP).
OpenAI, a prominent AI developer, has transitioned its commerce strategy from in-chat checkout to directing purchases through existing retailer applications, underscoring the necessity for underlying transaction infrastructure. This shift highlights that while AI can easily surface products, the primary challenge for retailers lies in downstream checkout processes, which were not built for machine efficiency, said Konstantin Klyagin, founder of QAwerk.
AI agents interact with digital systems differently than human users, making rapid, structured API calls that can inadvertently trigger rate limiting, bot detection mechanisms and disrupt session logic optimized for inconsistent human browsing patterns.
QAwerk identified a common failure where systems disagree on critical data points, such as pricing or inventory, or on the current state of a purchase. AI agents are unable to resolve these discrepancies without human intervention or judgment, Klyagin said.
Retailers should conduct load testing on their checkout APIs to ensure they can handle rapid, parallel calls from AI agents. They must also meticulously verify product data accuracy as it would be interpreted by a machine, according to Klyagin.
Furthermore, businesses need to confirm that refund and return processes clear correctly on the first attempt when initiated by an AI agent. Companies like Etsy, Glossier, Spanx and Vuori, which utilize platforms such as Shopify, are among those meeting these technical requirements.
Harley Finkelstein, president of Shopify, previously noted the increasing integration of AI across various aspects of e-commerce, including discovery and fulfillment.
The rise of AI commerce, supported by companies like Salesforce with its Agentforce Commerce, necessitates a re-evaluation of current e-commerce backend capabilities to prevent transaction failures and ensure a seamless customer experience.
Source: Search Engine Journal
