Over the past few months, the "AI will do your shopping for you" story has quietly changed. OpenAI rolled out Instant Checkout inside ChatGPT last year, letting people buy products without leaving the chat — then pulled it back this March after just a handful of Shopify merchants ever turned it on, and after Walmart found that purchases completed inside ChatGPT converted about three times worse than sending the same shopper to walmart.com. Meanwhile, Google is now rolling its Universal Cart out across Search and the Gemini app in the U.S. this summer, with YouTube and Gmail to follow later this year. Put those two facts together and you get a much more useful, much less hyped picture of where "AI shopping agents" actually stand right now — and it's not the picture most small business owners have in their heads.
The original pitch was that an AI agent would handle the entire purchase — find the product, compare options, pay, done — without a person ever landing on a retailer's website. In practice, that model has not taken off. Buyers didn't trust an AI window enough to hand over payment details, and the conversion data backed that up. OpenAI has since repositioned ChatGPT around product discovery rather than in-chat checkout, sending shoppers back to merchants' own sites to actually complete a purchase.
What's replacing it is simpler and, frankly, better for most businesses: an AI assistant helps someone figure out what to buy and where, then hands them off to that business's own website or booking page to finish the transaction. Google's Universal Cart, Perplexity's shopping features, and ChatGPT's product results are all converging on this same pattern. The AI is becoming a much smarter, much more conversational front door to search — not a replacement for your checkout, your booking form, or your sales process.
If you've been putting off AI shopping features because integrating with three different checkout protocols sounded like a project for a much bigger company, you can stop worrying about that. The businesses that will benefit from this shift aren't the ones with the fanciest agentic commerce integration — they're the ones an AI agent can actually find, understand, and recommend in the first place. That's a data and content problem, not an engineering problem.
AI shopping and research agents pull from the same places search engines always have: your website, your Google Business Profile, your product listings, review sites, and structured data on your pages. If your hours are wrong, your product catalog is outdated, or your pricing lives only in a PDF nobody's updated since spring, an AI agent won't call you to check — it will either guess, skip you, or quietly recommend a competitor with cleaner information. This is the same "garbage in, garbage out" problem AI has everywhere, just showing up in a new place: your sales funnel.
Open ChatGPT, Gemini, and Perplexity and ask each one plain questions a customer might ask: "Does [your business] offer X?" "What are their hours?" "How much does Y cost?" You'll learn fast whether these tools have accurate information about you, outdated information, or nothing at all. This costs fifteen minutes and tells you more than any theory about "agentic commerce" will.
Make sure your hours, address, pricing, and product or service details are accurate and say the same thing on your website, your Google Business Profile, and any marketplace listings. If you run an online store, keep your product feed and page content current — an AI agent recommending stale inventory or wrong prices is worse for you than not being recommended at all.
Since AI agents are now sending traffic to merchants' own sites rather than transacting themselves, the thing that actually determines whether that traffic turns into a sale is your existing checkout or booking process. A confusing cart, a broken mobile form, or a five-step scheduling page will lose that customer just as easily whether they arrived from an AI recommendation or a Google search. Spend your time there before you spend it on a shopping-agent integration.
Universal Cart, various agent payment standards, and similar integrations are still early, inconsistent, and mostly built with large retailers in mind. Unless you're already seeing meaningful traffic from AI tools and have a clear reason to integrate directly, there's little urgency here. Watch how these rollouts play out over the next few quarters and revisit the question once the winning approach is clearer — your time is better spent on the data-accuracy basics above.
The "AI will buy things for your customers" story turned out to be smaller than the headlines suggested — what's actually growing is AI as a research and recommendation layer that still sends people to your website or your door to finish the transaction. That's good news for small businesses: it means the competitive edge isn't a complicated integration project, it's the same fundamentals that have always mattered — accurate information, a clean product catalog, and a checkout or booking process that doesn't lose people at the last step. Get those right, and you're already prepared for wherever agentic commerce goes next.
DAOVA helps small businesses cut through AI hype like this and focus on the practical steps that actually move the needle. Explore AI Enablement.