Shopping online has quietly turned into a multi-tab endurance sport. You start with a vague idea—“something like this shirt,” “a cordless vacuum that doesn’t suck at sucking,” “a couch that actually fits my living room”—and 20 minutes later, you’re buried in listicles, affiliate blogs, and 14 open tabs that all say “best of 2026.” ChatGPT’s latest shopping update is basically a pushback against that chaos, turning the model into a more visual, more opinionated shopping companion instead of just another search box.

At the core of this update is a simple shift: instead of giving you advice and then sending you out to hunt down products alone, ChatGPT now pulls those products right into the conversation, and does it in a way that looks a lot closer to a modern shopping app than a wall of links. You can scroll through product cards, see photos, prices, specs, and reviews, and even get side‑by‑side comparison tables without ever leaving the chat. What used to be a multi-step journey—talk to an AI, open a new tab, search, filter, compare—becomes something you do in one place, in one flow.​

That shows up most clearly in “vibe-based” shopping—when you don’t have a product name, just a feeling. In OpenAI’s example, you upload a picture of a soft, floral short-sleeve shirt and say, “Find short-sleeve button-downs like this.” Previously, ChatGPT might have replied with style tips, brand suggestions, and maybe a few generic links. Now, it actually surfaces specific shirts that visually match the vibe: muted florals, relaxed fit, not too loud, wearable with jeans or chinos. You’re not just told “look at Abercrombie or Bonobos”—you see actual options laid out in a grid, with tags, images, and filters you can refine conversationally.​

The big quality-of-life win is the comparison layer. Instead of bouncing between retailer pages to compare fabrics, fits, and use cases, you get a structured table right in the chat: linen vs linen-blend, subtle floral vs bold leaf print, relaxed vs tailored-relaxed, everyday go‑to vs dressier casual. It’s the kind of comparison you’d normally build mentally after reading five product pages; ChatGPT now builds it for you. From there, it adds an opinionated layer on top—suggesting a “buy 1, add 1” strategy that turns endless browsing into a small, curated short-list.

Behind the scenes, all of this is powered by the Agentic Commerce Protocol (ACP), which OpenAI is extending to handle product discovery, not just checkout. ACP is essentially the connective tissue between merchants and ChatGPT: retailers share product feeds and promotions into this protocol so that when you ask for “mid‑range dishwashers that are quiet and fit in a small apartment kitchen,” ChatGPT isn’t guessing—it’s pulling from actual catalogs with up‑to‑date details. OpenAI says it’s optimized for speed, relevance, and coverage, meaning the results should feel fresher and more complete than the typical scraped product widget.​

Crucially, this isn’t just a theoretical ecosystem. OpenAI name‑checks big retailers like Target, Sephora, Nordstrom, Lowe’s, Best Buy, The Home Depot, and Wayfair as already integrated into ACP for discovery. If you’re on Shopify, your catalog is effectively already “open for business” in ChatGPT via Shopify Catalog, which feeds structured product data directly into the system with no extra work for individual merchants. That’s a big deal for small brands: suddenly, the same assistant people use to plan trips or write emails can also bring your products into highly targeted, high‑intent conversations.​

On the consumer side, this blurs the line between “research” and “shopping.” If you ask ChatGPT for, say, “a wrap dress I can wear to a beach wedding that’s under a certain budget,” it can now respond with a visual grid of options, filters for things like color or length, and a conversational loop where you refine based on what you like or don’t like. Upload a dress you saw on Instagram, and you can hunt for similar silhouettes or prints without knowing the brand name. The end result is that the AI starts to feel less like an advice columnist and more like a personal shopper that has a live feed into retail catalogs.

For merchants, this update is less about instant checkout and more about discovery and intent. OpenAI is explicitly stepping back from its earlier “Instant Checkout” push, saying it didn’t offer the flexibility they wanted. Instead, merchants can now route users to their own checkout flows, often through in‑app browsers where the brand’s site, design, and payment stack stay front and center. Shopify’s VP of Product calls this “AI shopping at scale”: millions of merchants discoverable through natural language queries, with purchases completed on their own storefronts rather than a generic one‑click layer.​

At the same time, OpenAI is leaving the door open for deeper, more “native” experiences via ChatGPT apps. Walmart is the headline example here, rolling out an in‑ChatGPT experience that connects discovery to a tailored Walmart environment. That includes account linking, loyalty integration, and Walmart payments, effectively turning ChatGPT into a front door into Walmart’s retail ecosystem—starting from “I need groceries for a week of cheap, healthy meals” and ending in a pre‑built cart you can check out inside a Walmart-branded flow. It’s available on the web today, with iOS and Android access coming soon.​

Taken together, you can see the shape of where OpenAI wants this to go: ACP as the foundation for “AI‑native commerce,” where things like personalization, local availability, and delivery ETAs can plug directly into the chat experience over time. If ChatGPT already knows your rough preferences, constraints, and context (“no gas stoves,” “small living room,” “no leather,” “within 2‑day delivery radius”), it can act as a high‑context recommender that’s aware of both you and real‑time inventory. The shopping flow stops being a one‑off search and becomes a relationship that gets better with every conversation.

Importantly, OpenAI frames this as an iterative build: they’re rolling these richer shopping experiences out to all ChatGPT free, Go, Plus, and Pro users over the coming week, and say they’re still learning from early launches and feedback. For now, the focus is squarely on improving product discovery—making it visual, fast, and grounded in real catalogs—before layering on more ambitious ideas. The bet is clear: if you can turn the assistant people already trust with their questions into a place where they also comfortably make purchase decisions, you don’t just improve shopping. You rewrite it around a conversation.