Why NaRestro
Most QR menus stop at showing the menu.
A photo of a paper menu behind a QR code isn’t a system — it’s a picture of one. Here’s what actually changes when the guest, the kitchen, the floor and the owner are all working off the same order.
| A typical QR menu | NaRestro | |
|---|---|---|
| What the guest sees | A PDF or photo of the paper menu, scanned and uploaded. | A live, searchable, filterable menu that takes the order itself. |
| AI-written content | Often generated live, per guest, and rarely disclosed as AI. | Written once by the owner’s own pipeline, always editable, and clearly labelled when it’s AI. |
| When an order is placed | Printed at a till, or relayed by a runner from memory. | On the kitchen display before the guest has put their phone down. |
| When food is ready | Someone has to notice, or the kitchen shouts across the room. | The assigned waiter’s phone buzzes the instant it’s plated. |
| Pricing model | Often a cut of every order, on top of a subscription. | One flat monthly fee. What you charge, you keep. |
| Ratings shown on the menu | Sometimes rounded up, or invented outright. | The restaurant’s real, current numbers — or nothing at all. |
What we won’t compromise on
Four principles, enforced in the code, not just the pitch.
Honesty is a feature, not a policy document
A generated dish photo says so, in the menu, where the guest is looking — not in a terms page nobody reads. A real kitchen photograph, once uploaded, is never quietly overwritten by a generated one. It is a rule enforced in the code that writes the menu, not a promise made in a pitch deck.
Generated once, not per guest
Dish stories, translations and narration are written and rendered ONE TIME, by the owner’s own content pipeline, and stored. Nothing on a guest’s screen is a live call to an AI model — which means no guest ever waits on one, no dinner rush is one API outage from a broken menu, and the bill doesn’t grow with how many people look at a dish.
A table is something you scan, not something you guess
Every table, every order, every staff session is scoped to a real, unguessable token issued by that restaurant — never a number a stranger could type into a URL. The kitchen, the floor and the guest each see only what their own token or session actually grants.
Built to be used, not demoed
Every feature on this site is running a real dining room today, not staged for a screenshot. Where something isn’t live yet — an integration, a payment provider — the product says so plainly, on the same screen an owner would look for it.
Proof, not a promise
Every page on this site describes software running a real kitchen tonight.
Latte Global Cafe, MVP Colony, Visakhapatnam — 201 dishes, 33 sections, three languages, one connected floor.