Skip to content
Victoria AI
Solutions · Point of sale

Customers ask right in the store, without waiting for anyone.

A screen on the floor, at the counter or by the shelves where customers ask by text or by voice and get products with a photo, price and dietary labels. No app, no queue.

It is built on your store's catalog, and the interface takes on its logo, its colors and its typeface.

Kiosk · idle
Kiosk attract screen: the logo and name of the store at the top, the headline "Ask me what to eat today", four suggested queries such as "something gluten-free" or "is there anything vegetarian?", the button "Touch to start" and the option to activate it by voice
What happens in the store

The questions all arrive at once, and always at peak hours.

None of these questions is hard. The problem is when they arrive: exactly when there is nobody free to answer them.

The same questions, all day long

"Do you have anything gluten-free?", "which one is vegan?", "where are the snacks?". The team answers the same thing dozens of times per shift.

Whoever is serving is busy

Every basic question pulls someone away from the register or the shelves. The customer waits, asks less than they wanted to, or leaves.

Restrictions that have to be known by heart

Gluten-free, vegan, sugar-free, low-sodium, keto. Filtering the menu or the catalog by hand depends on who is on the floor that day.

What is not available today

The customer asks for something that ran out. If whoever is serving does not have it in mind at that moment, the recommendation falls apart on its own.

This is what stores tell us when they ask about this. It is not a measurement of ours.

The query

Customers type the way they speak and get products back, not a search engine.

There are no categories to browse and no commands to learn. The customer types something like "something gluten-free for two" in everyday Spanish and the answer comes back with a photo, a price, a dietary label and how long it takes.

Kiosk · typing
A customer typing on the kiosk: the query "Do you have anything for celiacs?", the agent reply, the text box with the on-screen keyboard and three autocomplete suggestions below
Sample query.
Kiosk · the answer
The answer with products: two cards with a photo, name, price, dietary label and preparation time, three shortcuts to continue the conversation and the text box with the microphone
Sample products and prices.
  • Suggestions while typing — the keyboard proposes the store's most common queries, so someone who does not know what to ask still gets started
  • It follows the conversation — ask for more options, filter by price, compare two products or go back without starting over
  • Dietary labels from the catalog — gluten-free (sin TACC), vegan, sugar-free, low-sodium and keto come from the product record, not from a guess by the model
  • It says what is not available — products marked out of stock that day are left out of the recommendation instead of being offered anyway

The order is taken by someone from the store: the agent recommends and informs, it does not close the sale.

Suggestions panel

Alongside, it builds what might suit them.

On landscape screens the conversation takes the left side and on the right a suggestions panel is rebuilt with every answer: spicy, under a certain price, vegetarian. The customer sees the recommendation and the catalog at the same time.

  • Suggestions are rebuilt from the last thing asked, not from a fixed list
  • The notice of what is unavailable that day lives in the same panel, always in view
  • Availability by location, when the store has more than one point
Landscape screen · suggestions
The landscape view: the conversation on the left and, on the right, the suggestions panel with three products filtered by the last thing the customer asked for and a notice of what is out of stock that day
Sample products, prices and out-of-stock items.
Product view

Every detail without leaving the conversation.

Touching a product opens it up large: photos, the full description, attributes and price, with the conversation still open alongside. You go back without losing what had already been asked.

  • The main photo plus the secondary photos of the product
  • Attributes exactly as they are loaded: spicy, contains gluten, preparation time
  • The price, with whatever unit or portion applies

The same interface adapts to a vertical kiosk screen and to a landscape screen on the floor. It is the same conversation, not two different products.

Landscape screen · product view
The expanded product view: a large photo with more photos alongside, the full description, attributes such as spicy or contains gluten, the price, and the conversation still visible on the left
Sample product and price.
Voice mode

Customers speak and see the answer at the same time.

Speech recognition in Spanish, with no commands and no odd wording. While the agent answers out loud, the product cards appear on screen: nobody has to sit through a spoken list.

Voice mode · listening
Voice mode while listening: an audio waveform in the center, the query "something vegetarian to share" being transcribed below it and the exits to stop or switch to typing
Sample query.
Voice mode · answering
The answer in voice mode: the text of what the agent is saying on the left with its audio waveform, and on the right two product cards with a photo, price and preparation time
Sample products and prices.
  • You see what it understood

    the query is transcribed on screen while the customer speaks, so they correct it before the agent answers

  • Audio and screen together

    the spoken answer comes with the product cards showing photo, price and time

  • You can always type

    voice mode stops with one touch and the conversation carries on typed, without going back to the start

It works for the customer with their hands full, for the one who would rather talk and for the one who does not want to touch the screen. Typing is still one touch away.

Scope

What it does and what it does not do.

The in-store agent handles everything that happens before the customer decides. What comes after stays where it is today, and we would rather write that down here.

In the store

What the agent handles

Answering, recommending and showing. Everything that happens before the customer decides.

  • Availability — what is in each location and what is unavailable that day
  • Recommendation — based on what the customer asks for, with their restrictions
  • Photos, prices and descriptions — from the loaded catalog
  • Filters by attribute — gluten-free, vegan, sugar-free, low-sodium, keto
  • Text and voice in Spanish — on the same screen
Out of scope

What it does not do in the store

None of this goes through the screen on the floor: it stays where it is today, or it lives in another Victoria solution.

  • Closing sales or taking orders — the order is taken by a person
  • Processing payments — payment stays where it is today
  • Integration with the management system, POS or ERP — the store's catalog is loaded and maintained separately
  • CRM and customer management — a query on the kiosk does not create a lead
  • Digital channel — WhatsApp, social media and e-commerce are another solution

Several of the ones on the right Victoria already does elsewhere: the digital channel, the CRM and the integration with the management system. The platform is the same and they are switched on when the store needs them.

See the digital channel
Where it lives

The screen is already there. The customer just touches it.

No app, no download and no sign-up. The agent is installed on whatever mount works for the store and takes on its identity.

Kiosk at the entrance

A vertical screen standing at the entrance or next to the register. It is the one the customer who does not know what they want sees.

Tablet on the table or the counter

The same agent in a small format, for the floor, the bar or the service counter.

Screen on the shelf

Next to the products, for the question that comes up exactly when the customer is choosing.

With the face of the store

Logo, colors, typeface and tone of the brand. The agent answers the way the store answers, not the way a search engine does.

"Cantina Santa Rita", the store that appears on the screens on this page, is an example of customization: the interface is built with the brand of each store.

Getting started

From the menu to the screen.

It is a short project and it has an order. Whatever is not loaded, the agent does not offer: that is why the catalog comes first.

  1. 01

    The catalog

    The products with a photo, price, description and attributes: gluten-free, vegan, sugar-free, low-sodium, keto. That, and only that, is what the agent knows.

  2. 02

    The face and the tone

    The interface is adapted to the logo, the colors and the typeface of the brand, and we define how it speaks: more formal, more casual, more concise.

  3. 03

    A trial at a single point

    It is installed on one screen, we watch what customers actually ask and the answers are adjusted before adding the rest.

  4. 04

    The rest of the screens

    With the answers already tuned, it is replicated at the other points or locations, each with its own availability.

The catalog is maintained throughout the project. The out-of-stock items for the day are marked by the store on that catalog: the agent does not read stock from the management system.

How it is implemented
Frequently asked questions

What people always ask us.

Next step

Bring us your menu and we build it on that.

With your catalog, your photos and your prices we prepare the proposal: which screens, at which points in the store and with what scope.

It starts at one point in the store and, with the answers already tuned, it is replicated across the rest.

WhatsApp