Discovery Moves Into the Conversation
When a customer asks an AI assistant where to get an oat milk latte nearby, the answer is now a commercial surface, and Square wants its sellers to sit on it. On July 1 the company switched on a ChatGPT app and a Claude plugin that let eligible merchants get discovered and take orders inside those conversations. This is a quiet but consequential shift in where the storefront lives. For a decade, small food and beverage operators fought for visibility on maps, delivery marketplaces and social feeds. Square is betting that the next contested surface is the chat window, and that the merchants who show up there early will own an outsized share of the intent that AI assistants are starting to route.
The framing matters for enterprise buyers watching agentic commerce mature. Square is not selling a chatbot; it is turning its existing seller graph, the menus, hours, inventory and ordering flows, into structured data that AI assistants can read and act on. Morgan Kuntze, Global Partnerships Lead at Block, put the customer problem plainly, saying that consumer behaviors and preferences are constantly evolving and that business owners can easily find themselves playing an impossible game of catch-up. The pitch is that Square absorbs that complexity so a coffee shop does not have to build a separate integration for every model provider that reaches critical mass.
What Square Actually Shipped
The mechanics are deliberately boring, which is the point. Through the ChatGPT app, customers can browse a participating seller menu and place an order using Order by Cash App, with the transaction routed straight into the seller existing Square Online Ordering setup, including the point of sale and the Kitchen Display System. The Claude plugin offers the same discovery and ordering path inside Anthropic assistant. Crucially, eligible sellers are enrolled automatically. There is no API to build, no tools to configure and no separate app to maintain. Square syncs business information, menus and ordering data in real time, so the AI surfaces reflect what is actually available rather than a stale snapshot.
The first merchants to go live are United States food and beverage sellers with activated Square Online Ordering profiles, a segment where repeat, low-consideration purchases make conversational ordering a natural fit. Andrew Costaris, Digital Vice President at Partners Coffee, framed the upside in experiential terms, saying that what Square has built not only allows the team to continue offering analog, experiential moments, it creates more of them. That is the optimistic read: automate the transactional friction so staff can spend time on hospitality. The more sober read is that Square has made itself the connective tissue between thousands of small operators and a handful of very large AI platforms.
The Economics: No Commission, For Now
The commercial terms are the most aggressive part of the announcement. Orders placed through the AI channels carry no marketplace commission, no new contracts and no additional fees. For operators conditioned by years of delivery platforms skimming twenty to thirty percent off every ticket, a zero-commission channel is a genuinely different proposition. It also tells you where Square sees the value: not in taxing these early orders, but in keeping merchants inside its payments and software ecosystem as buying behavior migrates into AI assistants. The orders still settle through Square, and the seller still relies on Square hardware, reporting and payouts.
We would caution enterprise readers against reading zero-commission as a permanent state. Introductory economics are how platforms seed a channel before they meter it, and Square has every incentive to prove volume first and price later. The near-term value for a finance leader is measurement: order source is visible in Square reporting, so operators can actually see how much demand AI discovery drives before deciding what it is worth. That instrumentation is arguably more important than the fee waiver, because it lets a multi-unit operator quantify a channel that has, until now, been almost entirely invisible.
Voice, Protocols and the Standards Land Grab
Square is not treating chat as the only endpoint. The company is working with Amazon to bring sellers into Alexa+ experiences, extending the same discovery and ordering capability into voice commerce. It is also co-developing the Universal Commerce Protocol with Google, contributing a specification for local food ordering and delivery, and it sits on the AAIF Agentic Commerce Working Group and the W3C Web Payments Working Group. Taken together, this is a company trying to be present in every venue where an autonomous agent might complete a purchase, and trying to shape the rules of those venues rather than merely comply with them.
The standards angle is where CTOs should pay attention. Agentic commerce is currently fragmenting into competing protocols, with Google and its partners pushing the Universal Commerce Protocol and OpenAI advancing its own Agentic Commerce Protocol. A payments and software provider that participates in several of these efforts is hedging against a future where one protocol wins and strands the others. For merchants, the practical benefit is insulation: if Square abstracts the protocol layer, a seller does not have to bet on which standard prevails, in the same way it does not have to bet on which AI assistant its customers adopt.
What It Means for the Merchant Stack
For enterprise technology leaders, the Square move is a template for how incumbents defend distribution in the agentic era. The threat to any commerce platform is disintermediation: if an AI assistant becomes the place customers browse and buy, the platform that owns the payment and the back office risks being reduced to plumbing. Square answer is to make its plumbing the easiest way for merchants to appear inside those assistants, converting a disintermediation risk into a distribution advantage. The seller graph, the menu data and the ordering flows become assets precisely because agents need clean, structured, real-time inputs to act reliably.
There is a data-quality lesson here that generalizes well beyond food and beverage. Agents can only sell what they can accurately read, and the quality of a merchant structured content, its hours, availability, modifiers and pricing, now directly determines whether an assistant recommends it and completes the order. That is the same lesson large retailers learned this year as they poured effort into product feeds for AI discovery. Square has simply industrialized it for operators who will never build a data team, which is both the value of the offering and the reason it deepens their dependence on the platform.
The Risk of Renting Your Front Door
The strategic tension is unavoidable. By routing discovery through ChatGPT, Claude and Alexa+, Square is helping its sellers reach customers on surfaces that Square does not own and cannot fully control. If a model provider changes its ranking, restricts third-party ordering or launches a competing first-party experience, the merchants riding those integrations have little recourse. This is the classic platform-on-platform risk that delivery marketplaces created a decade ago, now reconstituted one layer up the stack, inside the assistants themselves. Zero commission does not neutralize that exposure; it simply makes it cheaper to accept.
Our read is that Square has made a smart, defensive bet that is also an admission of where power is concentrating. The company is right that its sellers cannot each negotiate with every AI platform, and right that aggregating them is valuable. But the long-term winners of agentic commerce may be the assistants that sit between customers and merchants, not the payment layer underneath. For now, Square has bought its sellers a seat at a table that is still being built, and given enterprise observers a clear, early case study in how commerce infrastructure adapts when the storefront becomes a conversation.
