This article digs into OpenAI’s shift away from its Instant Checkout feature. Now, the company’s building dedicated retail apps inside ChatGPT that send shoppers to retailers’ own sites to finish purchases.
It explores why OpenAI and retailers are rethinking their expectations. There’s plenty of friction with automated checkout, and companies like Google aren’t sitting still—they’re pushing ahead with their own AI shopping features.
You’ll also find some early signals about how shoppers behave, plus a look at how partnerships with merchants are changing. So what does all this mean for the future of AI-powered shopping? Let’s get into it.
OpenAI’s strategic pivot: from Instant Checkout to in-chat retail apps
ChatGPT is moving from a hands-off checkout tool to a platform that connects shoppers with retailers through dedicated apps. Instead of handling purchases entirely inside the chat, the new setup sends users to retailers’ own websites or in-app browsers to complete their orders.
This change gives merchants more control over checkout, data, and loyalty programs. OpenAI, on its end, is focusing on making search and product discovery better inside the bot.
The shift basically admits there’s a gap between what people want and what actually works when it comes to payments inside an AI chat. OpenAI has said it misjudged how tough it’d be to run smooth, end-to-end transactions, so now it’s putting product discovery and search relevance first.
What made Instant Checkout stumble
Several operational and data headaches slowed down the original plan:
- Onboarding and access were limited—just about 30 Shopify merchants were included, while Walmart has a catalog of around 200,000 products.
- Some sites were scraped to fill listings, so inventory, pricing, and shipping info sometimes ended up outdated or just plain wrong.
- Real-time syncing and consistent checkout experiences turned out to be trickier than expected, which frustrated both shoppers and merchants.
A new model for retailers: apps, not embedded checkout
Retailers like Shopify, Walmart, and Etsy are trying out ChatGPT apps or integrations. These let checkout happen on the retailer’s site, either in an in-app browser or a new tab, so merchants don’t have to build their own full-blown apps.
This makes it easier for retailers to join in, while still letting them control how checkout works and what happens after a sale.
With this app-based setup, retailers handle payments using their own systems. The AI helps with finding, comparing, and navigating products inside the chat, but doesn’t touch the actual checkout.
How the app approach benefits retailers
- More control over transaction integrity and customer data at checkout.
- Easier onboarding for merchants who aren’t ready to invest in complex AI checkout tools.
- Faster scaling across partner networks and different marketplaces.
- Better fit with loyalty programs and shipping rules that are already in place.
What shoppers do with AI shopping today
So far, most shoppers use AI tools for research, not instant purchases. The majority of transactions still happen after users are sent to retailer sites, where sellers manage pricing, stock, and shipping in real time.
Some quick stats:
- Only about 22% of surveyed users said they’d actually bought something inside an AI tool.
- Adobe/Semrush data show higher conversion rates when shoppers are redirected to retailer sites, which kind of proves how important real-time, accurate info is at checkout.
- Walmart said conversion rates were about three times higher for products sold directly on retailer sites via redirect than for things bought inside ChatGPT.
Industry dynamics and future prospects
Some analysts think this app-focused move could attract more partners, even big ones like Amazon. But it’s not a sure thing—Amazon’s been wary of web scraping and has taken legal action against startups before.
Meanwhile, Google has rolled out improvements like real-time product data, multi-item carts, and loyalty tie-ins. OpenAI still has some catching up to do there.
Across the board, OpenAI and retailers see AI-assisted shopping as a work in progress. It’s great for discovery and comparison, but not quite ready to replace traditional checkout. There’s a lot left to solve when it comes to data, trust, and making everything work together.
Conclusion: recalibrating expectations in agentic commerce
The shift toward in-app shopping apps feels like a practical adjustment in what AI can actually handle right now.
Most of the focus sits on discovery, personalization, and path-to-purchase inside the chat itself.
When it comes to checkout, people still prefer to finish up on retailer sites they already trust.
Maybe as data quality gets better and companies team up more, the AI shopping experience will feel smoother.
Here is the source article for this story: OpenAI’s first crack at online shopping stumbled. It’s preparing for the next wave