OpenAI is recalibrating its e-commerce strategy for ChatGPT. After launching direct purchasing capabilities last year, the company announced Tuesday it is scaling back the "Instant Checkout" feature. The move signals a shift from transactional execution to informational aggregation within the model's interface.

Initially, OpenAI positioned ChatGPT as a shopping assistant capable of handling carts and payments directly. However, internal metrics revealed low adoption. Users weren't converting conversations into purchases. Referral traffic data from October confirmed that vendors saw minimal revenue lift from the integration. Consequently, engineering efforts are now deprioritizing the standalone checkout module.

Instead, OpenAI is doubling down on product discovery. The company introduced its Agentic Commerce Protocol (ACP), developed alongside Stripe, to standardize how merchants expose data to the model. Rather than forcing a unified checkout flow, ChatGPT will act as an intermediary research layer. Merchants can build apps within the interface that route users to their own proprietary checkout systems.

This architecture allows for richer comparative metrics. Users will see side-by-side product images, pricing structures, and feature reviews without leaving the chat environment until the final purchase step. An OpenAI spokesperson noted the initial version lacked necessary flexibility. By handing checkout control back to merchants, OpenAI aims to improve the discovery engine while relying on established payment infrastructure elsewhere. For engineering teams watching the evolution of agentic workflows, this pivot highlights the current limitations of end-to-end transactional agents versus specialized discovery tools.

Source: TechCrunch