Somewhere between the hype and the eye-rolling, something real happened to online shopping: people started asking AI assistants what to buy. “Best moisturizer for dry skin under $40.” “Durable dog leash for a puller.” The assistant answers with actual products from actual stores, and yours is either in the answer or it isn’t. There is no ad budget that fixes not being in the answer, because these recommendations aren’t for sale.
The industry calls this agentic commerce, and the last year of it has been genuinely chaotic, so here’s the honest state of play and the practical checklist, in plain language, for regular-sized stores.
What’s actually real (and what already died)
A quick, honest inventory as of mid-2026:
- ChatGPT shopping is real as a discovery channel. Ask it a product question and it shows relevant products with images, prices, and links to buy on the merchant’s own site. Per OpenAI, results are organic and unsponsored, ranked on relevance. You cannot buy your way in; you can only deserve your way in.
- Buying inside the chat mostly isn’t. OpenAI’s much-hyped Instant Checkout was quietly removed in March 2026; Fast Company’s reporting found only a small handful of merchants ever went live. Meanwhile Google is pushing the other direction: its Universal Commerce Protocol launched in January 2026 with Walmart, Target, Shopify, Etsy and others aboard, and in-chat checkout has started rolling out through Gemini with major retailers. Translation: the plumbing is being fought over by giants, and it will keep changing.
- If you’re on Shopify, you’re already enrolled. Shopify’s Agentic Storefronts rolled out to merchants in early 2026: your catalog becomes discoverable in AI channels like ChatGPT by default, with channel toggles in your admin and orders attributed by AI source. Many store owners have this switched on and have never looked at it once.
The strategic takeaway: ignore the checkout wars, they’re not your fight. Your fight is discovery. Whoever wins the plumbing, the assistants will keep recommending products, and they can only recommend what they can read.
How a shopping agent sees your store
An AI agent doesn’t see your beautiful theme, your hero video, or your carefully art-directed lifestyle photography. It reads. It extracts: what is this product, who is it for, what does it cost, what’s it made of, when does it ship, what happens if the buyer wants to return it. Then it compares your answers against every other store’s answers and recommends whoever it could understand most confidently.
That reframes a decade of ecommerce advice. Vibes don’t parse. Specifics do.
The agent-readiness checklist
Everything below is doable in-house on a normal store. In rough priority order:
- Product titles that say what the thing is. “The Wanderer” is a lovely name and unreadable data. “The Wanderer: waterproof leather hiking boot, women’s” is both.
- Structured specifics on every product page. Materials, dimensions, ingredients, compatibility, fit guidance, use cases. Written as scannable facts, not buried in a paragraph of brand poetry. If a human customer service rep gets asked it weekly, it belongs on the page.
- Product structured data (schema). Product, price, availability, and review markup is how machines confirm what your pages mean. Shopify themes handle much of this automatically; verify yours does, and fix it if it doesn’t. Multiple analyses of AI shopping results find structured data heavily favoured, which makes sense: agents recommend what they can verify.
- Policies in plain sight and plain language. Shipping times, return windows, refund process. Agents check these before recommending, because their user will blame them for a bad experience. A store with a findable, humane return policy beats an identical store without one.
- Honest, answered FAQs. The objections people raise before buying, answered in two to four sentences each, on the relevant pages. You’re writing the script the assistant will use to reassure your customer.
- Consistent pricing everywhere. Agents compare your site, your feeds, and marketplaces. A price mismatch between your collection page and product page, or your site and your feed, reads as unreliability and gets an otherwise-winning product skipped.
- Let the right robots in. Check your robots.txt: the retrieval crawlers that power live AI answers (OpenAI’s search bot, PerplexityBot, Google’s) need access, or you’re invisible by your own hand. Blocking training crawlers is a separate, legitimate choice; just don’t block the ones doing the recommending. Some platforms have added default restrictions, so this is worth two minutes of actually looking.
- Reviews, still and always. Assistants summarize review sentiment when recommending. The store with 400 honest reviews beats the store with 12, in machine conversations exactly like human ones.
The five-minute audit
Before optimizing anything, find out where you stand. Open the major assistants and shop for your own products the way a stranger would: “best [your category] under [your price point],” “[your category] for [your niche’s specific need].” Three outcomes:
- You show up, accurately. Congratulations, protect it: keep data clean and prices consistent.
- You show up, wrongly. Stale prices or dead products usually trace to an old feed, a forgotten sales channel, or crawlers reading pages you’ve abandoned. Fix the source.
- You don’t show up. Work the checklist above, and also the broader visibility playbook, because product recommendations lean on the same trust signals as everything else: reviews, mentions, and being describable. Our plain-language AI search guide covers that half.
Repeat quarterly. This landscape reorganized itself twice while we were writing this post.
The honest outlook
Nobody, including the companies building it, knows exactly what shape AI shopping settles into. What’s already true: discovery is moving into conversations, the recommendations are earned rather than bought, and the stores that win them are the ones whose products are legible, specific, honestly priced, and well reviewed. Which, you’ll notice, is just being a good store with the receipts visible. The robots, it turns out, shop like your most careful customer.
If you’d like help finding out how your store reads to them, our Brand Audit now covers AI shopping visibility alongside everything else. The checklist above, though, is yours free. Go get recommended.