Product presentation
Interactive AI avatars for public spaces
A digital employee that answers visitors out loud around the clock, handles routine requests and performs actions in your systems — without ever leaving your perimeter.
Most effective where visitor load is high, staffing is short and wayfinding is hard.
01What it is
Your digital employee
The avatar hears the visitor and answers out loud, the way a person would — it holds eye contact and knows the schedule, the programme, the floor plan, the rules and the prices of your venue specifically, because it is trained on your materials and fed a knowledge base you keep current.
Three form factors, one knowledge base
Kiosk on the floor
A terminal with a screen, microphone and speakers. Can run with no internet at all.
Widget on your site
The same character and the same knowledge base, in the visitor browser.
Module in your app
Embeds into the venue existing mobile application.
The knowledge base and the character are shared across all three. Data does not leave the venue perimeter and no external APIs are called. Integration options are agreed separately.
02Conversation as the interface
The avatar takes the visitor all the way to the outcome
We assemble a set of actions in your systems, specific to your venue.
Booking
A screening, a class, a guided tour, an appointment.
Sale
A ticket through the connected point-of-sale register.
Status
Of an application, a payment, a flight, a document.
Intake
A complaint, an incident report.
Escalation to a person
With the conversation context already collected.
Sample dialogues
The visitor speaks in their own words. The avatar answers from the venue knowledge base and, where needed, performs an action.
03What the venue gets
Five effects you see in the first season
Takes load off the front line
Routine questions go to the avatar so your staff can work the cases that actually need a person.
Runs 24/7 with no rota and no churn
A new season does not mean hiring and training another shift.
Never has an off day
Just as courteous in the two-thousandth conversation of the day as in the first.
Shows what people actually ask
De-identified request analytics — what people look for, what they fail to find, where they go.
Doubles as a talking point
The technology is still rare, so the installation draws attention and ends up in visitor social feeds.
04Timeline
Three weeks to a working avatar
In three weeks the avatar answers from your knowledge base on our infrastructure. Terminal delivery and integration with your venue services run in parallel — their timing depends on your systems.
Launch in two waves
The avatar goes live on the floor
It answers visitor questions from the venue knowledge base. That alone is enough for the installation to start working and producing data.
Actions come online
Booking, sale, status, intake — as each of your systems becomes ready, without taking the avatar offline.
05Content
The knowledge base is your own documents
The programme, the floor plans, the regulations, the price list, the schedule. We do not ask you to write anything from scratch.
Updated from a browser, with no involvement from us
Answers strictly from the base — the avatar does not make things up
Forbidden topics are agreed before launch
Conversation history is available in the administrator log
06Delivery options
Three options, one platform
Turnkey terminal
- What we deliver
- A finished terminal: enclosure, screen, audio path, server, platform.
- Hardware
- Entirely our responsibility.
- Best fit
- You want a fast launch and a single party accountable for the result.
Software on your hardware
- What we deliver
- The platform, the deployment, the ongoing support.
- Hardware
- Procured by the venue to our specification.
- Best fit
- You already have terminals, info stands or a server fleet.
Hybrid
- What we deliver
- Terminal or software at the venue, the language model on our servers in Russia.
- Hardware
- Procured by the venue to a lighter specification.
- Best fit
- You want top model quality at a low cost of entry.
All three give you the same platform and the same level of support. The choice comes down to whether the venue already has hardware, how fast you need to launch, and how demanding you are about language model quality: a fully isolated perimeter running a top-tier model needs a markedly more expensive build, and the hybrid removes that constraint.
07Track record
Where this already runs
Government customers in the Gulf
An isolated perimeter, an internal certificate authority, integration with the Nafath national citizen identity system. Arabic, including regional dialect.
Rostov region, a school museum
An avatar of the nineteenth-century mayor of Rostov-on-Don. Russian language, an audience of children.
Our own platform
A cloud system with autoscaling GPU pools runs on the same codebase as the on-site installations. In production for over a year.
Team
Eleven engineers and specialists, together since 2022
We started in medical computer vision and 3D reconstruction, with publications at IEEE and MICCAI. Today we specialise in real-time AI systems inside closed perimeters: generative face and speech models, infrastructure, security for on-premise deployments, animation and 3D graphics.
We build and maintain the entire stack ourselves — from the server to the character on the screen.