Square made Squaremates generally available to U.S. sellers on November 12 — a free AI agent that now serves as the front door to the Square Dashboard for more than 4.5 million sellers worldwide. It analyses sales trends, drafts marketing campaigns, optimises staffing schedules, and builds supplier orders — multi-step operational work that used to mean a spreadsheet, a late night, and a lot of guesswork. But the more significant move isn't the agent itself. It's that Square simultaneously launched an MCP server and CLI that let sellers plug any AI tool directly into their Square account.
The business tools small and mid-sized operators already pay for are quietly becoming AI platforms. We've seen it with Xero launching XeroForce for accounting workflows and eXp Realty adopting an MCP-first approach for its 87,000 agents. Square's move is the clearest signal yet that this pattern has reached the point-of-sale layer — the software that touches every transaction a small business runs.
What Square actually launched
Squaremates is the rebrand and expansion of Managerbot, which entered open beta in April 2026. Each seller's agent now runs in its own secure workspace with access to cloud compute, which is what lets it handle longer, multi-step jobs rather than simple Q&A.
The practical capabilities matter more than the marketing language. Square's examples include: analysing a period of sales to identify lapsed customers and generating a win-back campaign or discount; reviewing a seller's entire catalogue and making bulk updates to prices, descriptions, channel visibility, and images; weighing labour costs against sales and returning a cost-optimised staffing schedule for review. Crucially, the agent now produces rich artifacts — a polished graphical sales breakdown or a fully designed email campaign rather than a paragraph of text in a chat window.
"Running a business is an endless string of judgment calls that are all on the owner. How to price a new menu item, whether to add a shift, what to do about a slow week," said Willem Avé, Square's Global Head of Intelligence. "We built Squaremates to help share that load."
Two features are still in progress. A restocking capability, currently in beta, checks what's running low, prepares an order, and builds it on the supplier's own site for the seller to complete. An inbox-aware daily briefing is listed as coming soon.
The open platform play is what matters
The agent itself is competitive but not unique. Shopify's Sidekick, Toast's IQ assistant, and Intuit Intelligence all offer free AI capabilities within their platforms. What separates Square is the decision to open the account to outside tools.
Square MCP lets sellers connect a third-party AI assistant — Claude, ChatGPT, Cursor, or any MCP-compatible tool — to their Square account once, then ask questions and make changes from within that assistant. The Square MCP server is hosted at mcp.squareup.com/mcp with OAuth sign-in, or sellers can run it locally via npx square-mcp-server start.
Square CLI goes further, giving developers and technically minded sellers command-line access to Square's APIs. Combined with AI coding tools, this enables Squaremates Apps — custom applications that Square hosts with a live connection to the seller's business data.
Donnie McClanahan, who runs four cafes across Knoxville, Tennessee, built a labour board app with the CLI and an AI coding agent that pulls schedules and timecards from all locations into a single live view. "A lot of the morning used to be playing catch-up just figuring out who's where and whether every location is covered," McClanahan said.
This open approach aligns with where the industry is heading. According to Postman's 2026 State of the API report, 56% of organisations are now experimenting with or using MCP in production, with 24% already in production. Theory Ventures' AI in Practice Survey found MCP usage rose from 33% to 57% of teams year-over-year, while the competing A2A protocol fell from 17% to 8%.
What this means for small business owners
The practical implications split into three tiers depending on how you run your business.
If Square is your primary platform, Squaremates is worth turning on immediately for analysis and drafting work. As TWIMCO's review noted, the best fit is "an owner-operator with one to a few locations who already runs everything through Square and does the analysis in their head, or not at all." Start with questions — sales patterns, lapsed customers, slow-moving inventory. Keep prices, discounts, published schedules, and catalogue changes behind approval until you've built confidence in its accuracy.
If Square is one system among several, proceed with more caution. Squaremates only sees the data that flows through Square. If your inventory lives in a separate tool or your online store runs on another platform, the agent will give confident answers based on half the picture. That's a common limitation across platform-native AI — it's not unique to Square, but it's worth naming clearly.
If you're technically inclined or work with someone who is, the MCP and CLI tools are the genuinely exciting part. Building a custom app that solves a problem specific to your business — and having Square host it with a live connection to your data — creates real operational value. But as TWIMCO cautioned, both are still labelled beta in Square's developer documentation. Treat connecting an outside assistant as a project, not a toggle.
One important caveat: Squaremates is currently available only to U.S. sellers, and not to those on Square's Multi-Account Suite. Square operates in Australia, but there's no announced timeline for Squaremates to reach this market.
What to watch
FintechBits identified the strategic logic clearly: the real retention play isn't the chat agent — it's the custom apps sellers build on Square's infrastructure. A seller who creates a labour board or a tip-splitting tool on Square has a reason to stay that a free assistant never provides. That's retention through custom software, and it's a much stronger moat than any chatbot.
Four signals will determine whether this matters beyond the press release. First, whether Block publishes adoption or retention figures in its fourth-quarter earnings — Square has shared no usage data since saying "thousands" had used Managerbot during the beta. Second, whether the MCP and CLI shed their beta labels. Third, whether the restocking and email briefing features ship, since those are what connect Squaremates to the world beyond Square. Fourth, when — or if — Squaremates reaches international markets including Australia.
For Australian business owners watching from the sidelines, the pattern is more important than the specific product. The tools you already use for payments, accounting, and operations are racing to become the AI layer for your business. The Grout Guy built an AI workforce on Servly that cut work orders from two days to two minutes. Xero launched a no-code agent builder. Now Square is doing the same for point-of-sale. The question isn't whether your business tools will gain AI capabilities. It's whether you'll be ready to use them when they arrive.
Sources
- Square Expands its Agentic Platform with Squaremates and Powerful New AI Tools — Square Press Release
- Square takes Squaremates AI agent out of beta — SiliconANGLE
- Square's AI takes on menu, labor, marketing tasks for restaurant operators — Fast Casual
- Squaremates review: a useful back-office agent for Square sellers — TWIMCO
- Squaremates agent: a strong free AI tool after 198 days — FintechBits
- Square Statistics 2026: GPV, Sellers, Take Rate — Axis Intelligence
- One in Four Organizations Now Run AI Agents on APIs in Production — Postman / BusinessWire
- AI in Practice Survey 2026 — Theory Ventures
