On July 1, 2026, Square switched on a ChatGPT app and a Claude plugin that let customers find a restaurant, read the menu, and place an order without ever touching the restaurant’s website. According to Square’s press release, eligible US Food and Beverage sellers with an active Square Online Ordering profile were opted in automatically, with no API to build, no new tool to configure, and no added marketplace commission. If you sell on Square, you may already be taking AI-assisted orders and not know it.
The interesting part is not the integration. It is who did the work.
What Square actually shipped
Per the July 1 announcement, orders placed inside ChatGPT or Claude use Order by Cash App and route straight into the seller’s existing Square Online Ordering setup, including the POS and the Kitchen Display System. The order source shows up in Square’s reporting, so an operator can see how much volume the AI channel is producing.
Square says it handles the plumbing: syncing business details, menu data, hours, and ordering information in real time, with discoverability controlled from the existing Square Dashboard. Square also states it is working with Amazon on Alexa+ voice ordering and co-developing the Universal Commerce Protocol spec for local food ordering with Google.
Two numbers from the announcement are worth writing down. Square cites NielsenIQ research that more than 42% of consumers now use AI tools for shopping tasks such as discovery, comparison, and selection. It also cites a Morgan Stanley outlook projecting that agentic shoppers could drive close to $385 billion in US e-commerce spending by 2030. Both come from third parties rather than Square’s own measurements, and the 2030 number is a forecast, not a result.
The fee math is doing the persuading
Square is not charging extra marketplace commission on these orders. A restaurant pays its normal Square online processing rate, which Square’s pricing lists as 3.3% plus 30 cents on the free plan and 2.9% plus 30 cents on paid Square Online tiers after the January 13, 2026 pricing change.
Compare that with delivery marketplaces. DoorDash’s own merchant pricing page lists three partnership plans at 15%, 25%, and 30% commission, with pickup orders at 6% across all plans.
So a $40 ticket costs a restaurant about $1.46 to $1.62 in processing through Square’s AI channel, versus $6 to $12 in commission on a DoorDash partnership plan. That is the pitch, and it is a strong one. The obvious caveat: the marketplace is also supplying the courier, the demand, and the customer service. An AI assistant sending someone to your existing pickup and delivery setup is not the same product.
What to do: if you are a Square F&B seller, open your Square Dashboard reporting and filter by order source this week. You now have a channel with a cost of sale in the low single digits. If it is producing orders, that changes how you think about your delivery mix.
Your platform is now your AI distribution layer
Square is not alone in this. Shopify announced Agentic Storefronts in its Winter ’26 Edition, describing one setup in the merchant admin that syndicates products into AI conversations across platforms including ChatGPT, Perplexity, and Microsoft Copilot, with orders flowing back into the admin with AI channel attribution. Google has been pushing the same idea from the other end, which we covered when Google’s universal cart went live across Search and Gemini.
The pattern is consistent. The platform negotiates the integration, normalizes your catalog, and turns you on by default. You inherit the channel.
That is great if you are on Square, Shopify, or another platform doing the work. It is a problem if you are not. There is no plugin that negotiates a spot in ChatGPT’s app directory for a self-hosted store. Merchants on WooCommerce, Magento, or custom stacks are competing against catalogs that are already parsed, standardized, and available to agents.
What to do if nobody is opting you in
You cannot conjure a platform partnership. You can make your store readable to the agents that do crawl and cite it.
- Fix your structured data first. Product, Offer, availability, price, and LocalBusiness hours need to be accurate and machine-readable. Our guide to structured data written for LLMs covers the markup patterns that matter here.
- Publish policies as plain text. Returns, shipping windows, allergen or ingredient notes, service areas. Agents answering customer questions need somewhere to read the answer.
- Keep one canonical price and stock source. Conflicting prices across your site, feed, and third-party listings is the fastest way to get skipped or misquoted.
- Decide what you allow. Check your robots rules and WAF settings so that legitimate assistant crawlers are not blocked while abusive scrapers are. Store owners are already dealing with the volume problem we wrote about in why storefronts became the internet’s top bot target.
- Instrument it. Add referral tracking so you can tell an assistant-driven session apart from organic search. If you cannot measure the channel, you cannot argue for investing in it.
The parts to be skeptical about
Automatic enrollment cuts both ways. Your menu, hours, and prices get republished inside an interface you do not control, and the conversation about your brand happens where you cannot see it. Square lets sellers manage discoverability from the dashboard, which is worth reviewing rather than assuming.
Scope is also narrow right now. This launch covers US Square F&B sellers with online ordering active. Retail, services, and non-US sellers are not in it, and Square describes further AI surfaces as coming rather than shipped.
Then there is the customer relationship. Marketplaces got criticized for owning the diner. An assistant that answers questions, compares three restaurants, and places the order sits in the same spot. Lower fees, same intermediary problem. Running your own assistant on-site is one counterweight, and the numbers are better than most expect, as covered in our look at stores running their own AI agents. If you deploy one, disclosure rules now apply, which we covered in the AI chatbot disclosure deadline.
The takeaway
Choosing a commerce platform used to be a decision about features and fees. In 2026 it is also a decision about who negotiates your distribution. Square just opted millions of sellers into two of the largest assistant surfaces for free. Shopify did something similar earlier this year. Nobody is doing it for the self-hosted store down the street.
Check whether your platform is doing this work for you. If it is, go read your channel reporting. If it is not, clean up your product data now, because being legible to agents is the only version of this you control.
Sources
- Square press release, “Square Introduces New ChatGPT and Claude Integrations, Helping Sellers Reach Customers Through AI-Powered Discovery,” July 1, 2026. Available at squareup.com
- Square, agentic commerce overview. Available at squareup.com/us/en/ai/agentic-commerce
- NielsenIQ, “42% of consumers now use AI tools to shop, NIQ data shows”. Available at nielseniq.com
- Morgan Stanley, agentic commerce market impact outlook. Available at morganstanley.com
- DoorDash merchant pricing (Basic 15%, Plus 25%, Premier 30%, pickup 6%). Available at merchants.doordash.com
- Shopify Newsroom, “Introducing Shopify Agentic Storefronts”. Available at shopify.com/news
- Shopify Help Center, agentic storefronts documentation. Available at help.shopify.com
Last reviewed: July 29, 2026.
Disclosure: this site may earn a commission from some links on this page, at no extra cost to you. That never affects which tools or platforms we cover or what we say about them.
