I saw the future of retail, and its all AI

Several people are gathered around a bleach blond man in a bright pink suit suspended in a clear plastic tube. With a microphone in front of him and a giant sign reading TALK TO ME placed above, “Mike” waits, hands clasped patiently in front of his body, to take questions from his public. “Mike” is a hologram, powered by ChatGPT and created by a company called Hypervsn.

The responses “Mike” gives to audience comments and questions are on a three-ish second delay, but the stunted flow of conversation might not matter much — an attendant at the Hypervsn booth tells me that when “Mike” and his ilk are deployed to stores, they are meant to act more as icebreakers, a way to lure potential customers into engaging with a brand. (The world’s largest holographic display thrills tourists at the Las Vegas Sphere and was also made by Hypervsn.)

“Mike” is one of the flashier products exhibited at the National Retail Federation’s enormous annual trade show, where more than a thousand retail-adjacent companies descend on New York City to try to sell each other products and services that will then be used to sell actual goods to the rest of us. There are the big names (Google, Alibaba, Amazon) along with scores of other companies that help power the online and in-person shopping experiences but are unknown to the average consumer. No matter what vendors are exhibiting, though, one thing is clear: The future they imagine for all of us is overwhelmingly filled with AI — often in ways that a consumer would probably hate.

A few feet away from “Mike” is another hologram, also suspended midair in a clear box a few feet tall; this one is a small, gnome-like creature dressed in a monochromatic, vaguely medieval outfit (I did not catch his name). At times he responds in rhyming verse to audience questions, his tiny hands gesturing as naturally as a theme park animatronic. Retail clients are increasingly requesting non-human holographic characters, a Hypervsn attendant tells me, in an effort to distance themselves from fears that AI will replace or eliminate people’s jobs. If this is the future of retail, even “Mike” may not have a place in it.

There were many promises made at the National Retail Federation’s massive trade show, held in mid-January. Some exhibitors swore to provide “touchscreens for everything,” others reassured convention-goers that “commerce favors the bold.” But the most urgent hopes were around AI. Shopify announced that “the commerce renaissance is here” (italics theirs). Other exhibitors touted “smart people counting” and “AI customer flow analysis.” We come to this place not for magic, but for “merchandizing execution with AI.”

Much of this is likely word salad to the average consumer — but if you shop online, the AI creep has become inescapable. Retail and tech companies have crammed AI into just about every step of the purchasing process, from designing products, to discovery and comparison, to calling brick and mortar stores, to trying on clothing, and finally checking out. At NRF’s trade show, Google announced an open-source standard called the Universal Commerce Protocol (UCP) that allows retailers and AI agents to communicate and integrate, so shoppers can buy products from Target within Google’s AI Mode, for example, without going to Target’s website. As tech companies build more direct buying features into chatbots, some are hoping coupons and deals can entice purchases: Google also announced that retailers will be able to set up discounts for shoppers browsing in AI Mode. Monetizing AI-powered shopping is clearly a priority for Google: the company had a large presence at the show, and CEO Sundar Pichai gave a keynote address (on the other hand, Pichai had no public appearances at CES, and Google had no major announcements).

As AI has snaked its way into every gadget, platform, and service imaginable, it’s hard not to wonder who the intended user is — and whether anyone, ever, has wanted the kind of features AI makes possible.

Pizza chain Papa Johns is one of the partners integrating Google’s agentic shopping features into its ordering system. Customers will soon be able to order via a chatbot that can autofill in their last order, suggest items based on dietary restrictions, add coupons and deals, and use a “pizza assistant” to get recommendations for group orders. In a demo video shared by Papa Johns, the chatbot assistant asks for the number of people in the group. Don’t want to count how many people are eating? No problem — just upload a photo of everyone in the group into the Papa Johns chatbot.

There is also, of course, the evolution of industries that were upended by the rapid introduction of AI. Retailers used to prioritize tweaking their websites to get Google to place their link above competitors’ in search — but now search engine optimization, or SEO, has branched off into new AI-focused acronyms: AEO (answer engine optimization), GEO (generative engine optimization), GSO (generative search optimization), and likely other future terms the general public will come to despise. A company called Fabric promises to help brands and retailers monitor their performance within platforms like ChatGPT, Gemini, and Perplexity, measuring how often your items are surfaced relative to your competitors. A representative from Fabric says some retailers are especially good at getting surfaced in AI chat — Nordstrom, for example — but acknowledged that nobody really knows how the systems work or what they choose to elevate. Fabric’s AI monitoring service starts at $500 a month, according to its website.

Many of the products and services on display at the NRF show are on the back-end — firms working on things like inventory management tools or logistics, where consumers have less insight. But the consumer is an ever-deepening well of data; the promise of AI in retail is that these tools can extract more and more information from us, which can be used to sell us more things.

Solum is a South Korean company specializing in digital displays and shelves — those signs in the supermarket playing glossy ads for chips or beer, for example. But the next iteration is to hyper-personalize and target shoppers in stores, the way digital retailers are able to do to online shoppers. Using a technology built by a South Korean startup called SpaceVision, a retailer could assign a number to each shopper and track their movements through a store: how long they watched an ad, whether they simply walked by without engaging, if they picked up a product off the shelf. The store could then use that data to trigger certain behaviors: for example, if you watched a certain chip ad on a display, the store could push through a deal for you for beer to go with your chip purchase; it applies the tracking of online shopping into the real world. As I approached a wall of Coors and Heineken (not affiliated with SpaceVision) cans arranged on a shelf at the Solum booth, I noticed a screen above the display. It took me a moment to realize my own image was staring back at me, captured by the SpaceVision demo. With a bold box surrounding my face, I was Global ID number 485, gender: female, age group: 18–29. Retailers can use the data to track the demographic breakdown of people who pass displays, along with their “glance” and “attention” rates and conversions.

A SpaceVision employee working the booth at the show said that when deployed in the real world, video is deleted a millisecond after the metadata is captured — they are focused on milking the engagement data that can be used to retarget shoppers later on. The technology has so far only been used in select stores in South Korea and Japan, but the staffer acknowledges that shoppers in Europe or the US might be uneasy about the technology becoming part of the in-person shopping experience. It creates, essentially, a vast data dragnet of IRL shopping habits to mine for new and more invasive ways to sell things — the monetization of that omnipresent surveilling eye. Shoppers in Europe would balk at cameras. On the other hand, the staffer says, “Asian people don’t give a fuck.”

Equapack’s booth at NRF 2026 stood out for a few reasons: For one, it was empty when I stumbled upon it in one of the trade show’s many, many rooms. It was also calm: No giant LED screens, no holographic gnomes, no humanoid talking robots, and no grand AI-related promises. The booth consisted of several long, white shelves with dozens of shopping bags, totes, cooler bags, and other retailer packaging carefully lined up. Among the selection were iconic and instantly recognizable names: the classic red and white Supreme streetwear logo on plain white bags; US Open drawstring backpacks; a tote bearing the logo of Broadway’s The Lion King. Unlike everything else I saw during the day, Equapack’s offerings were designed to be closer to the human at the end of the transaction — not to add additional layers of data, AI-powered analysis, and abstraction.

The trade show used to be more tactile like this, says Eran Rothschild, Equapack’s founder and CEO. All the packaging they design starts from a problem that the brand needs to solve: perhaps they need something to keep perishable items cold for a certain number of hours, or maybe they simply want a bag that is aesthetically beautiful and evocative. Equapack does not promise AI-powered solutions because it doesn’t use AI, though Rothschild admits that it probably should, at least for back-end operations — design is just 25 percent of the job.

“We’ll probably never use AI visualization because it’s not true to the product that we’re going to be delivering,” Rothschild says. “We prefer making samples and making that tactility. No one’s going to purchase something that they can’t see.”

AI slop has only one goal: scale. It does not matter if the AI-generated clips flooding social media are good or even entertaining — they simply need to take up space, and by extension, a human’s time. Could an AI chatbot on every online ordering page transform the customer experience, whether they’re ordering clothing or pizza delivery? Maybe. But it also feels like added bulk that has little bearing on whether the thing we bought — the actual product coming into our homes — is better than it would have otherwise been.

As we’re talking, I realize that I personally own several bags that Rothschild has worked on, including packaging from The RealReal, a secondhand clothing and accessories marketplace. Whether the item you purchase is $30 or $3,000, The RealReal mails products in fabric dustbags, a practice largely limited to luxury retailers. I reuse every dustbag I get, carefully storing items in them to protect them and extend their life. For every project, Rothschild thinks about how the consumer will experience the packaging — is it reusable, is it luxurious? Does it make the purchased product feel like money well spent? Of all the tech and services I saw throughout the day, Equapack’s offerings are one of the few that a consumer will actually hold in their hands and take home with them. There is intention, a sense of intimacy. I will likely never think back fondly to an experience chatting with an AI shopping assistant. The humble shopping bag, though, will linger.