Case Study · Hospitality

Sushi On Fire

A luxury restaurant run end to end by AI.

Industry

Hospitality

What We Built

Voice ordering, chat, one-tap checkout, owner CMS

Status

Live working product

Stack

React · Gemini Live · Function Tools

Sushi On Fire is a demonstration restaurant created by Agentify AI. The checkout is a simulated Apple Pay sheet that declares itself a demonstration, and every screen below is the actual product as it runs today.

The Problem

The menu is beautiful. It cannot take an order.

Restaurant sites are brochures: pretty, static, and mute. They show the food, then hand the hungry guest a phone number. This one had to take an order three ways, by voice, by chat, and by tap, and make the paying part disappear.

Say a roll out loud and it lands in the cart. Confirm with your face and the kitchen has an order number.

The Solution

What Agentify AI Built

One restaurant, three personas: a guest who orders by voice, chat, or tap; an owner who runs the house from a browser; a chef who runs the line from a ticket board.

1. The Agents

Flame speaks over Gemini Live and the chatbot types over the same function tools: add to cart, check out, book a table. Every confirmation comes from a real tool result, never a claimed one.

2. The Commerce

A cart that survives reload and a path from menu to paid in about four taps: a simulated Apple Pay sheet with a Face ID confirm, and no typing anywhere.

3. The Kitchen

A ticket board for the chef, modeled on merchant tablets: what is new, what is on the line, what is ready. One tap advances a ticket, and a chime sounds when an order lands.

“Added 1 Rainbow Roll to your cart. Your total is twenty dollars and fifty-seven cents. Ready to check out?”

Flame, the voice concierge, acting on a spoken order through a real tool call

Inside the Product

The Actual Interface

Captured from the live product as it runs today: the shipped version sits beside the reimagined one, and every screen is the real interface, not a mockup.

The first build of Sushi On Fire beside the reimagined version: the same restaurant with a different standard of light, type, and appetite

Shipped vs Reimagined · before left, after right

The first build beside the elevation: same restaurant, a different standard of light, type, and appetite.

A grid of sushi dishes photographed under one locked art direction: matte slate, charcoal ground, one warm rim light

One Photographic Signature · 25 dishes, generated photography

Every dish reshot by AI under one locked art direction, so the whole menu reads as a single photographer's work.

The simulated Apple Pay sheet on the Sushi On Fire checkout, asking for a Face ID confirm

Cart to Paid, One Tap · simulated Apple Pay, disclosed as a demo

The black button opens the sheet, the sheet asks for a face, and the order number arrives. No forms on the fast path.

Flame the voice concierge and the text chatbot side by side, both acting on the same cart through the same function tools

Two Agents, One Tool Belt · Gemini Live voice, Gemini chat

Flame hears you; the chatbot reads you. Both act through the same functions against the same cart.

The Sushi On Fire dining room film breathing smoke and ember behind the hero section

The Films · both loops began as stills

The dining room breathes smoke and ember behind the hero; the blade ignites in the craft section. Both films were generated from the site's own photographs.

The Sushi On Fire owner admin in the browser, showing orders and reservations as they arrive

The House CMS · every order, every table request

Orders, reservations, menu, and copy all land in an owner admin that lives in the browser.

The chef's ticket board with new, on the line, and ready lanes, ticket numbers, and urgency clocks

The Kitchen Line · modeled on merchant tablets

Ticket lanes with daily numbers and urgency clocks: new, on the line, and ready, each order one tap from the next stage.

The kitchen ticket board stacked into swimlanes on a phone screen

The Chef's Pocket · live tickets, quantity-first type

The same board stacks into swimlanes on a phone, so the line keeps moving even when the tablet is across the kitchen.

A confirmed Sushi On Fire order with a real order number and pickup window, placed without typing a word

The AI Moment

Confirmed with a face, not a form.

The whole argument lives in one screen: a guest who never typed a word holds a confirmed order with a real number, a pickup window, and a receipt trail in the owner CMS. Friction is the thing being demonstrated, by its absence.

4

Taps, Menu to Paid

0

Forms on the Fast Path

The Film

The Blade, the Signature

A still photograph taught to burn: the craft film behind the brand.

Poster frame of the Sushi On Fire craft film: the chef's knife igniting along its blade

Tap to play · produced by Agentify AI from the site's own stills

The Outcomes

What Changed

27

Photographs, One Signature

Every dish and atmosphere generated under one locked art direction, so the whole menu reads as a single photographer's work.

3

Personas, One Product

A guest who orders by voice, chat, or tap; an owner with a CMS in the browser; a chef with a live ticket board.

4

Taps, Menu to Paid

The black button opens the sheet, the sheet asks for a face, and the order number arrives. No forms on the fast path.

Say it, tap it, or type it. The kitchen hears all three.

Your menu could take its own order.

We will walk through Sushi On Fire live, then map what voice, chat, and one-tap ordering look like on your menu, in your house.