OpenAI’s ambitious move into AI-powered shopping is facing early resistance, as consumer behaviour proves harder to disrupt than expected.

The company recently partnered with major retailers, including Walmart, Etsy, and Shopify, to introduce agentic shopping, a model designed to allow users to discover and purchase products directly within ChatGPT. The initiative, anchored by a feature known as Instant Checkout, aimed to remove traditional e-commerce friction by enabling transactions without leaving the chat interface.

At launch, the concept signalled a potential shift in how consumers interact with online retail. Walmart alone made up to 200,000 products available through the system, positioning the experiment as one of the most significant attempts yet to merge conversational AI with commerce.

However, early performance has fallen short of expectations.

Despite the convenience offered, sales generated through ChatGPT lagged behind traditional e-commerce channels. Conversion rates remained weak, highlighting a disconnect between the technology’s capabilities and established consumer habits.

A key limitation was structural. The initial system supported only single-item purchases, while typical shopping behaviour involves browsing, comparing, and purchasing multiple products in one session. This restriction reduced the appeal of the experience, even for users already comfortable interacting with AI tools.

The outcome underscores a broader reality in digital commerce: shopping is not purely transactional. Consumers value the process as much as the outcome, from comparing options to building carts and maintaining a sense of control over decisions.

ALSO WATCH MARKETING EDGE ONTV

In response, retailers are beginning to recalibrate their strategies.

Walmart, for instance, is shifting focus away from completing purchases entirely within ChatGPT. Instead, the company is expanding its in-house AI assistant, Sparky, integrating it across platforms such as ChatGPT and Google Gemini to support, rather than replace, the shopping journey.

The revised approach blends AI assistance with familiar e-commerce structures. Features under development include synchronised shopping carts, multi-item purchasing, and more personalised recommendations, allowing consumers to retain control while benefiting from AI guidance.

Early indicators suggest this hybrid model may be more effective. According to available data, users interacting with Walmart’s AI assistant spend approximately 35 per cent more per order, pointing to AI’s growing influence on purchasing decisions, even if it does not directly complete transactions.

The shift reflects a wider industry trend.

Technology companies are increasingly exploring how AI can enhance commerce without fully automating it. Google is advancing its own framework through the Universal Commerce Protocol, while Amazon continues to develop AI-driven shopping tools within its ecosystem. Microsoft, meanwhile, has begun integrating purchasing capabilities into its Copilot platform.

While the long-term potential of AI in commerce remains significant, early results suggest its role will evolve more gradually than initially anticipated.

For now, AI appears better suited to improving product discovery, recommendations, and personalisation rather than replacing traditional purchasing behaviour entirely.

The experiment with agentic shopping highlights a critical lesson for the industry: innovation in commerce must align with how consumers actually behave, not just what technology makes possible.

As retailers refine their approaches, the future of AI in shopping is likely to be defined less by automation and more by augmentation, supporting consumers through the buying process rather than attempting to control it.