Home / AI News / Products

Square launches a conversational AI ordering kiosk

Square's new touchscreen kiosk includes a voice AI layer that lets customers talk through their orders.

Products2026-10-141 min readBy PickAI
Square launches a conversational AI ordering kiosk – PickAI illustration

Square has launched new restaurant and retail hardware: a large touchscreen ordering kiosk that customers can also talk to. It is called Square Kiosk, and its main feature is a conversational AI voice layer.

What it does

Square describes the voice system as agentic and conversational. It is designed to understand a seller's full menu, listen and respond in real time, and let customers order the way they would naturally speak to staff. During the conversation, customers can ask questions about items and request modifications or substitutions.

The pill-shaped kiosk has a 21.5-inch touchscreen and uses presence detection to greet people as they approach. Payment happens on the device through a contactless reader or card slot.

Pricing

Square Kiosk costs $1,999 and is available to businesses on Square Plus or Premium plans. There is also a software fee of $99 per device per month.

Why it matters for builders

Voice ordering is one of the clearest real-world tests of AI agents: the domain is narrow, the menu is known, and success is easy to measure. Square bundling it into mainstream point-of-sale hardware suggests the technology has crossed from pilot projects to off-the-shelf product.

For developers, the interesting design problems are reliability and grounding. A voice agent must never invent menu items or prices, must handle noise and interruptions, and must confirm orders clearly before charging. The usual approach is to tie the model tightly to structured menu data and to show the order on screen as it is built, so the customer can catch mistakes.

The pricing model is also worth noting: hardware plus a monthly software fee per device. If you are building voice or agent products for small businesses, recurring per-location pricing is becoming a familiar expectation.

Builders exploring their own voice experiences can look at real-time speech models and voice tools such as ElevenLabs and GPT-Live 1 in our directory.

Source: The Verge. This article is PickAI’s own summary and analysis based on the original reporting.

Related tools on PickAI

↗

ElevenLabs

elevenlabs.io

Realistic AI voices, text-to-speech and voice cloning.

↗

OpenAI API

platform.openai.com

API access to OpenAI's GPT, image and audio models.

↗

GPT-Live 1

developers.openai.com

OpenAI full-duplex voice model for live conversations while an agent reasons in the background.

More AI news

All news →
The Verge

ChatGPT Chrome extension can read your open tabs

OpenAI upgraded the ChatGPT Chrome extension so it can use your open browser tabs when you tag Chrome, and ChatGPT Work can control tabs for tasks like filling forms. Atlas is gone; this brings similar browser help into Chrome.

2026-10-15
Manus

Manus 1.5 speeds up and builds full-stack apps

Manus launched version 1.5 with faster task completion, higher quality, and full-stack web app building with auth, databases and deploy. Manus-1.5-Lite is free for all users; full Manus-1.5 is for subscribers.

2026-10-16
Meta AI Research

Meta releases Muse Spark Open weights

Meta released Muse Spark Open, an 816B-parameter open-weight multimodal model (42B active) with a 1M-token context window for coding and agents. Weights are on Hugging Face; hosted APIs include OpenRouter, Groq and others, and it can run on two Apple M5 Ultra Mac Studios via MLX.

2026-10-15

Get the weekly AI tools drop

New tools and releases for builders, once a week. No spam, unsubscribe anytime.