Square's New Kiosk Lets Customers Order by Talking to an AI

Square has announced a new self-service ordering kiosk with a built-in conversational AI voice layer, marking the payments company's most direct push yet into agentic ordering hardware. The device, called the Square Kiosk, pairs a 21.5-inch touchscreen with software the company says can hold a real-time conversation about a restaurant's menu.

The announcement matters because it moves voice-based AI ordering from experimental pilot territory onto standard point-of-sale hardware sold to Square's broad base of food sellers. If the conversational layer works as described, it changes the counter experience for thousands of businesses; if it does not, Square has bolted a voice assistant onto an increasingly crowded kiosk market.

What exactly did Square launch?

The Square Kiosk is a large, pill-shaped countertop terminal designed for customers to place their own orders. Its headline feature is what Square's press materials call an "agentic, conversational voice layer." The phrase "agentic" here means software that takes actions — building and modifying an order — rather than only answering questions.

Square says the voice layer "fully understands a seller's menu and listens and responds in real time, enabling customers to order the way they'd naturally talk to staff." That is the company's claim, and it carries the full weight of the usual speech-recognition caveat: menus are full of brand-specific names, sizes, and modifiers, and vendors rarely publish error rates for such systems.

While talking with the AI agent, customers can ask questions and request modifications and substitutions, according to Square. The touchscreen remains the fallback and the primary display: it measures 21.5 inches diagonally, a size comparable to a desktop monitor, large enough for a full menu layout alongside the conversation.

The device also includes presence detection. Square says the Kiosk "can welcome you as you approach," meaning sensors or a camera register a customer approaching and trigger a greeting on screen. The company did not specify in the announcement materials which sensing technology it uses or how the data it captures is handled — a gap worth noting in any hardware that watches for approaching people.

The takeaway: this is a first-party Square product, not a third-party add-on, which means the ordering hardware and the AI conversation are sold as one integrated system.

Why does an AI voice layer on a kiosk matter?

Self-service kiosks have spread across quick-service restaurants over the past decade, but almost all of them rely on tapping through nested menu screens. Square's bet is that voice removes that friction: a customer says what they want, the way they would to a cashier, and the order builds itself.

The claim rests on two hard problems the announcement does not fully address. First, real-time speech understanding in noisy restaurant environments remains difficult; a counter near an espresso machine or a fryer is an acoustically hostile place. Second, "fully understands a seller's menu" implies structured knowledge of every item, modifier, and substitution rule — the kind of thing that today often requires manual configuration or per-menu training.

Square has an advantage here that smaller kiosk vendors lack: its sellers already manage their menus inside Square's point-of-sale software. If the Kiosk reads that structured menu data directly, the conversational layer has a clean source of truth. The company's announcement does not detail the technical pipeline, so the mechanics remain Square's word against the demo.

The takeaway: the product's promise is real-time natural-language ordering, and the burden of proof sits on how it performs in loud, busy stores rather than in a press release.

What can customers actually do at the Kiosk?

According to Square's announcement, the interaction model includes:

• Greeting customers on approach via presence detection
• Answering questions about menu items through the conversational voice layer
• Accepting modifications and substitutions spoken in natural language
• Displaying the full menu and order on the 21.5-inch touchscreen

Square frames the experience as ordering "the way they'd naturally talk to staff" — its words. That framing invites the obvious comparison test: can the agent handle the improvised, half-finished sentences real customers actually use? Square did not publish benchmarks, transcripts, or third-party evaluations alongside the launch, so there is no independent evidence yet.

The takeaway: the feature list is conversational ordering end to end, but Square has offered no performance data to verify fluency or accuracy.

How does this fit the wider market?

Square operates one of the largest small-business point-of-sale ecosystems, and kiosks have become a proven revenue lever for quick-service restaurants, where they tend to raise average order sizes and reduce labor pressure at the counter. Adding AI voice to that form factor puts Square in direct competition with both legacy kiosk manufacturers and a wave of voice-AI ordering startups.

The launch also reflects a broader industry pattern: payments and POS companies layering generative AI onto existing hardware rather than building standalone AI products. For Square's sellers, the practical question is pricing, availability, and setup burden — none of which the announcement materials detailed. Square has not said what the Kiosk will cost, when it ships, or whether the voice features require a separate subscription.

Those omissions are the story's soft spot. A kiosk with a working voice agent at the right price could reshape counter operations for small restaurants; the same kiosk with unreliable speech recognition would push customers back to the touchscreen — which Square, conveniently, still includes.

The takeaway: Square has the distribution to make conversational kiosks mainstream, but cost and real-world reliability details will decide adoption.