The impact of AI on buyer journeys is undeniable.
Recent data reveal that 63% of European consumers use answer engines for shopping research and to compare brands, models, or prices.
Still, according to the Channel Engine’s Marketplace Shopping Behavior 2026 report, only 17% of surveyed marketplace shoppers in the US, the UK, France, Germany, and the Netherlands feel comfortable completing purchases with AI.
In light of these figures, OpenAI’s decision to abandon Instant Checkout, as well as eBay’s policy change to ban unauthorized buy-for-me bots from crawling its platform, appear to be sound business moves, given the low adoption rate of the fully automated checkout process.
Despite this, AI commerce is growing incrementally, even though a lack of trust, control, and confidence in outcomes holds back its adoption.
Let’s take a closer look at why the Instant Checkout feature was pulled back from ChatGPT and how agentic commerce can affect shopping on online marketplaces in the coming years.
Key takeaways
- OpenAI has removed the Instant Checkout functionality from ChatGPT in early March 2026.
- The early adopters of the Instant Checkout feature included Walmart, Shopify, and Etsy.
- In November 2025, eBay’s CEO Jamie Iannone hinted that eBay may join the Instant Checkout program in the future.
- In February 2026, eBay changed its User Agreement to ban unauthorized use of third-party buy-for-me agents on its marketplaces.
- Research reveals that consumer trust in AI declines when it comes to prefilling shopping carts, completing checkout, or automatically reordering products.
What is agentic commerce, and how does it affect consumer behavior?
Agentic commerce, a fully automated shopping process managed by AI agents that review quality sources to help consumers discover, compare, negotiate, and buy products, is still in its infancy.
The term was first used in early 2025, shortly after Stripe released the Agentic Commerce Protocol in partnership with OpenAI.
It is used to denote generative AI platforms that generate tailored suggestions while understanding customer intent to help shoppers purchase products from reliable retailers within the AI application’s interface without ever visiting the retailer’s site or listing on an online marketplace.
ChatGPT’s Instant Checkout was the first agentic commerce agent available to consumers in the US, capable of finding products on Shopify, Etsy, or Walmart.
Even though Amazon’s Rufus or Alibaba’s Qwen can complete purchases for shoppers within the platform, Instant Checkout was the first agent to provide horizontal coverage by allowing ChatGPT users to purchase items from different sellers across the web.
However, the statistics show low adoption of agentic commerce among consumers, though recent research has shown that 51% of shoppers in the US would feel comfortable allowing AI to manage the entire online shopping process, including checkout.
On the other hand, an overwhelming majority of retailers are open to agentic commerce and see it as a strategic opportunity.
A glance at the Instant Checkout and the Agentic Commerce Protocol
OpenAI released the Instant Checkout feature for ChatGPT on September 29, 2025, after developing the Agentic Commerce Protocol (ACP) in collaboration with Stripe, which powered the feature.
Instant Checkout initially allowed ChatGPT users in the US to purchase items from Etsy. Still, only a few days later, Shopify unlocked the feature for millions of merchants, enabling them to fully automate the checkout process.
By the end of October, PayPal announced its adoption of the ACP, revealing plans to support AI-powered checkouts, while Instacart launched an app within ChatGPT that enabled end-to-end Instant Checkout in early December 2025.
The new feature allowed users to search for products, view product details, and complete purchases in a few minutes, without leaving ChatGPT’s interface, by using the Agentic Commerce Protocol to ensure the safety of all transactions.
The ACP generated a checkout inline within the chat when a user was ready to make a purchase. Stripe would issue a Shared Payment Token and pass it to ChatGPT via an API to allow it to complete the transaction while protecting the user’s payment details.
However, only 8% of ChatGPT’s adult users in the US had tried Instant Checkout within the first month of its release, and usage rates remained low throughout the five-month trial period.
The end of OpenAI’s Instant Checkout feature
Only months after launching its agentic commerce tool, OpenAI announced its decision to abandon it. On March 5, 2026, the company acknowledged that the highly publicized feature would be discontinued because it wasn’t generating conversions.
Only a dozen Shopify merchants have integrated the feature into their checkout processes, while consumers who used ChatGPT to research and compare products largely preferred to complete purchases on the retailer’s website.
Still, canceling the Instant Checkout doesn’t mean OpenAI plans to abandon its ecommerce plans. Instead, the company intends to route purchases through third-party apps available in ChatGPT.
Consequently, online shoppers will still be able to buy products from within ChatGPT’s interface, but only through dedicated apps, allowing retailers to manage the payment processing.
ChatGPT will still use the Agentic Commerce Protocol to allow third-party apps to function within its interface, but the protocol will no longer serve as a one-click buy engine.
Reasons OpenAI’s checkout experiment didn’t work
Low adoption, customer satisfaction (or rather dissatisfaction), and conversion rates are only a part of the reason OpenAI pulled back the Instant Checkout feature.
The premise that shoppers would purchase an item directly from a thread after researching a product has proven inaccurate, as most consumers don’t shop this way.
So, for the Instant Checkout feature to attract significant attention, consumers would have to change their behavior, as brand loyalty or return history are still more important to most shoppers than convenience.
However, the reasons OpenAI’s experiment didn’t deliver the expected results are much broader than shopper behavior. The feature didn’t include a range of functionalities, such as inserting promotional codes or adding multiple items to shopping carts, which may have contributed to its low adoption rates.
Even though ChatGPT is widely used as an answer engine, OpenAI’s attempt to become a one-stop ecommerce solution hasn’t worked out due to limited access to up-to-date information for product catalogues, inventory, and updates.
Hence, OpenAI’s attempt to automate checkout within an answer engine shows that horizontal agents that rely on external product catalogues are much less successful than vertical agents.
Buy-for-me chatbots, AI shopping assistants, and agentic commerce on online marketplaces
The beginning of 2026 has been important for the growth of the agentic commerce marketplace.
On January 11, Google announced the launch of the Universal Commerce Protocol (UCP), developed in collaboration with Shopify, Etsy, Target, Walmart, and Wayfair, and backed by Mastercard, Visa, Stripe, and more than twenty other companies.
The UCP was updated in mid-March to improve the real-time management of inventory updates and enable multi-item checkouts. In doing so, Google addressed the very issues that plagued OpenAI’s Instant Checkout.
Meanwhile, eBay changed its User Agreement in late February to prevent unauthorized buy-for-me bots from completing purchases on its marketplaces, in an attempt to protect its auction dynamics and minimize the risk of fraud.
Furthermore, as of March 2026, Amazon is still blocking ChatGPT from showing its product listings in search results despite agreeing to a $50 billion partnership spanning the next eight years.
At the same time, millions of shoppers are using Amazon’s shopping assistant Rufus AI to discover products on the platform and complete purchases, suggesting that access to the online marketplace’s vast product inventory may be crucial to the agentic commerce tool’s success.
The difference between horizontal and vertical AI agents
ChatGPT’s Instant Checkout was a general-purpose shopping interface that attempted to provide a checkout solution for thousands of online stores at once.
Google’s Native Checkout feature is a similar type of horizontal agent, but the major difference is that the Google Pay ecosystem is already in place, whereas OpenAI’s horizontal agent failed because it required each merchant to create a payment bridge between ChatGPT and its payment processing system.
Although they make product discovery much easier, horizontal agents often struggle to provide accurate tax or shipping information to users because they have access only to external product inventories.
On the other hand, vertical agents like Amazon’s Rufus or Walmart’s Sparky are much more reliable and consequently have much higher usage rates as they have access to the platform’s internal inventories, which enables them to provide users with a high level of accuracy during the checkout process, while protecting their payment data.
How should online sellers prepare for the future of agentic commerce?
The statistics clearly show that, for now, the majority of consumers are comfortable using answer engines to learn about a product or a category of products, and with AI making suggestions on which item to purchase, as long as they complete the purchase.
However, shoppers are still slow to adopt a fully automated checkout process because they don’t fully trust AI to buy products for them.
Still, recent research suggests that by the end of the decade, the agentic commerce market could reach up to $5 trillion worldwide, indicating rapid growth in customer engagement.
That’s why Generative Engine Optimization of product pages remains the best way for online sellers to prepare for the future of ecommerce.
Are you struggling to optimize your product listings for discovery on eBay? Schedule a call to discover how Webinterpret can improve the visibility of your listings across eBay marketplaces.
About Webinterpret
Webinterpret supports merchants selling on eBay.
Our AI-based solutions enable more effective selling through automated listing localisation, optimisation, advertising, and returns, and ensure all products placed on EU markets are GPSR-compliant.
By giving your international and domestic customers a full, end-to-end local shopping experience, Webinterpret improves your conversion rates and helps establish your business locally and globally.
