May 1, 2026
Boarding2Now/Depositphotos.com
Are Retail Apps Within ChatGPT and Claude in Trouble, Or Is it Too Early To Tell?
With OpenAI having discontinued (or scaled back, depending upon who you ask) its much-vaunted Instant Checkout feature in March — in favor of moving the purchase journey through retailer-specific app, instead of the general ChatGPT platform itself, the future of agentic AI experiences tied to consumer buys appears to remain in flux.
The new paradigm? ChatGPT apps and Claude “connectors,” which are essentially the same thing under a different name. Retailers can, and are, creating their own apps for shopper use, leaving discovery up to the AI platform and handling checkout through their own connected app. As Modern Retail’s Mitchell Parton pointed out, even though a full 10% of all apps on ChatGPT are shopping-oriented, including those from Target and Walmart, conversion rates may not be as rosy.
Parton quoted Alpic chief of staff Dimitri Ewald on that subject. Ewald’s company, which builds and distributes apps for both of the aforementioned AI platforms, says conversations with clients have been revealing.
“For the moment, to be honest, adoption and conversion are pretty low. People don’t even know that there are apps in the ChatGPT store,” Ewald said.
And the low conversion rates may not be news to those following the previous Instant Checkout functionality, either.
“Several people involved with Instant Checkout told Modern Retail and other outlets that the program didn’t drive sales and that some merchants didn’t want OpenAI overseeing the checkout process,” Parton wrote.
Retail Apps May Require Evolution Before Hype Matches Promise (If it Ever Does)
Ewald wasn’t the only expert cited by Parton who was taking a wait-and-see approach to the current environment regarding AI sales funnels. Jason Goldberg, chief commerce strategy officer for Publicis Groupe, took a middle-of-the-road stance, pointing to the potential for evolution versus the existing status quo.
“I don’t think there’s anything wrong with a retailer or brand doing an app, and I think there are benefits to be had; there are learnings to be had, and data to be collected, and muscle memory to be built by brands and retailers doing it. I wouldn’t discourage someone from doing it. I don’t think it’s economically very exciting or material. In the long run, unless it evolves dramatically, I don’t think it’s going to be an important part of the agentic commerce ecosystem,” Goldberg said.
“I would not say, ‘Apps are never going to work, and they’re dead.’ What I would say is, if they do work and they’re important to shoppers, they’ll probably look pretty different than the ones we have today,” he added.
The reasons provided for retailers and brands clambering for AI partnerships and integrations were many — from FOMO to wanting to set anchors in early in hopes that the situation would improve. Or, simply to control the information being put in front of shoppers as AI adoption improves, both on the consumer side and on the corporate side.
Other problems are in evidence as well, according to both brands and retailers as well as OpenAI itself. Issues surrounding app approval (and attendant red tape), bugs in the code, and a dearth of usage data are mixed with a general anxiety about handing off customer information to the models.
For its part, OpenAI remains optimistic about ironing these problems out as things move forward.
“We’re still early in building this out, and we recognize there are areas where the developer experience needs to improve,” a spokesperson said, indicating that the company aims to make ChatGPT “more reliable, more predictable and easier to build on over time.”
ChatGPT, Claude: A Useful ‘Low-Cost Focus Group’ or Merely ‘Experiments To Drive Press Releases’ for Brands and Retailers?
Goldberg would go on to describe that the learning value attached to experimentation within the agentic AI ecosystem could be valuable to retailers, one which could translate to greater leverage.
“You could think of launching a retail app on ChatGPT as a low-expense focus group that you can then use to improve the agents you might build for your own vertical website,” he said.
But Juozas Kaziukėnas, an independent e-commerce analyst, was much more critical, making comparisons between AI app launches and the eventually disastrous NFT launches which now constitute more of a memory than a lasting store of value — a popularity contest, then, rather than a lasting marker.
“There’s no visible success story of any of these apps being a meaningful driver for any of these companies. We can theorize how useful these apps could be or can’t be, but I think, for now, it’s quite clear they’re just experiments meant to drive press releases and don’t really impact anything at all,” he said.
BrainTrust
Nicholas Morine
Recent Discussions
Discussion Questions
Is a lack of concrete proof of conversions and a pullback from instant checkout a serious problem for retail / AI platform partnerships? Is the hype yet to be realized, or misplaced?
Where does your opinion fall on the “low-cost focus group” versus “experiments to drive focus groups” positioning as regards retail and brands moving into AI apps and search? Are both too cynical, or no?
What does the next evolution of agentic AI in retail look like to you?
Poll
Loading …
The current proof is still an early proof. AI is changing and growing so quickly that anything that seems perfect today (Is there such a thing as perfect?) will be outdated in six months. Until that hockey-stick-like improvement flattens a bit, whatever seems great today will be a disappointment in the near future. The hype is warranted, but brands and solution providers should be careful not to overdo it.
I had dinner recently with some very senior executives who think that AI preferences are changing in a matter of week. It’s just too early to declare victory.
Maybe I missed the point. Consumer are and will continue to increase the use of Agentic AI to make online purchases. Once the consumer has found the item and the desired ‘seller’, s/he will already be at the ‘seller’s’ website. At that point, the consumer checks out and pays. Why then, is it necessary for the AI agent apps to provide instant checkout? The only reason I can see is to avoid the possibility that this is a first purchase from that particular seller, and the consumer will have to provide a payment method, whereas the Agentic AI app will probably already know that information and could apply it automatically. It is not a given that the consumer will want to always use the same payment method, however. Bottom line? I won’t miss instant checkout on an Agentic AI app.
The AI landscape is still evolving, so it is too early to say for certain how commerce will reshape itself. That said, it was always obvious that AI agents were only going to be one channel among many for shopping because consumers don’t consolidate their buying activity in one space. And what’s currently happening is that AI is growing as a discovery channel, but not so much as a buying channel. We see this in our own data: among youth consumers 45.3% have used AI for ideation and discovery when shopping; but only 9.1% have used it to actually transact. That means retailers should focus more on optimizing for the former and drive through traffic to their own transaction mechanisms.
It’s early in to expect retail/AI partnerships to function as a valuable tool for retail shoppers or to glean useable customer relationship data. But small and narrow engagement trials to see where resources should be placed should be a constant exercise. For example, currently there appears to be small but effective traction in AI app usage in grocery retail back of house cost savings and efficiencies, in ordering and supply chain. This is showing value right now.
It feels too early to call. Awareness is still low, so weak adoption and conversion aren’t all that meaningful yet. Until more people actually know these experiences exist and start using them, it’s hard to say what their impact will be.
I agree with pretty much everyone here: much too early to tell (would give it until Tuesday, at least 🙂 )
“Too early to tell” is fair, but incomplete. Low conversion is not just a maturity problem; it is a structural signal. Instant Checkout did not convert, and merchants were reluctant to let the platform own the customer relationship. That reluctance was rational. Walmart’s experience makes the point precisely: products completed inside ChatGPT converted at roughly one-third the rate of those redirected to Walmart’s own site. The discovery layer and the transaction layer do not automatically move together, and retailers who understand that distinction are working to preserve independence at the layer that matters economically.
The NFT comparison is too cynical. NFTs had no utility. AI agents actually do something for the shopper. The smarter takeaway: retail apps inside AI platforms are useful tuition, but no substitute for building agent capabilities you own and control.
More Discussions
Claire’s Comeback: Will its ASMR-Driven Campaign Aimed at Gen Alpha Succeed?
Will Claire’s sensory-driven campaign be the first step to a successful comeback?How Can Home Improvement and Furnishings Retailers Work Around Gen Z’s Lack of Homeownership?
How can home improvement and furnishings retailers re-orient themselves to serve an up-and-coming demographic?Will Subway’s First-Ever Value Menu Impress Guests (Or Is it Too Little, Too Late)?
Is Subway’s first-ever value menu a big play, or too little, too late?
The current proof is still an early proof. AI is changing and growing so quickly that anything that seems perfect today (Is there such a thing as perfect?) will be outdated in six months. Until that hockey-stick-like improvement flattens a bit, whatever seems great today will be a disappointment in the near future. The hype is warranted, but brands and solution providers should be careful not to overdo it.
I had dinner recently with some very senior executives who think that AI preferences are changing in a matter of week. It’s just too early to declare victory.
Maybe I missed the point. Consumer are and will continue to increase the use of Agentic AI to make online purchases. Once the consumer has found the item and the desired ‘seller’, s/he will already be at the ‘seller’s’ website. At that point, the consumer checks out and pays. Why then, is it necessary for the AI agent apps to provide instant checkout? The only reason I can see is to avoid the possibility that this is a first purchase from that particular seller, and the consumer will have to provide a payment method, whereas the Agentic AI app will probably already know that information and could apply it automatically. It is not a given that the consumer will want to always use the same payment method, however. Bottom line? I won’t miss instant checkout on an Agentic AI app.
The AI landscape is still evolving, so it is too early to say for certain how commerce will reshape itself. That said, it was always obvious that AI agents were only going to be one channel among many for shopping because consumers don’t consolidate their buying activity in one space. And what’s currently happening is that AI is growing as a discovery channel, but not so much as a buying channel. We see this in our own data: among youth consumers 45.3% have used AI for ideation and discovery when shopping; but only 9.1% have used it to actually transact. That means retailers should focus more on optimizing for the former and drive through traffic to their own transaction mechanisms.
It’s early in to expect retail/AI partnerships to function as a valuable tool for retail shoppers or to glean useable customer relationship data. But small and narrow engagement trials to see where resources should be placed should be a constant exercise. For example, currently there appears to be small but effective traction in AI app usage in grocery retail back of house cost savings and efficiencies, in ordering and supply chain. This is showing value right now.
It feels too early to call. Awareness is still low, so weak adoption and conversion aren’t all that meaningful yet. Until more people actually know these experiences exist and start using them, it’s hard to say what their impact will be.
I agree with pretty much everyone here: much too early to tell (would give it until Tuesday, at least 🙂 )
“Too early to tell” is fair, but incomplete. Low conversion is not just a maturity problem; it is a structural signal. Instant Checkout did not convert, and merchants were reluctant to let the platform own the customer relationship. That reluctance was rational. Walmart’s experience makes the point precisely: products completed inside ChatGPT converted at roughly one-third the rate of those redirected to Walmart’s own site. The discovery layer and the transaction layer do not automatically move together, and retailers who understand that distinction are working to preserve independence at the layer that matters economically.
The NFT comparison is too cynical. NFTs had no utility. AI agents actually do something for the shopper. The smarter takeaway: retail apps inside AI platforms are useful tuition, but no substitute for building agent capabilities you own and control.