The store run by AI.
One AI is the manager, the cashier, the guard and the marketing team. Customers scan, shop and walk out. A 21 m² store, craned onto site and trading in under an hour, under any brand.
The staffed store
stopped adding up.
Thousands of forecourts, campuses, hospital sites and small towns have real demand. But a clerk on every shift sinks the numbers before the first basket is scanned, so SmartNovo removes the line items that break the model.
Labour breaks the model
Staff are half to two-thirds of a convenience store's running cost. Poland's minimum wage nearly doubled in five years.
Closed when demand peaks
Most small-format stores shut by 9pm and on holidays, exactly when an always-on format captures the demand that walks away.
Theft keeps climbing
Shrinkage runs 1.5 to 2% of European retail sales, and small stores rarely have real-time loss prevention.
Slow to open, expensive to leave
A conventional fit-out means a lease, a build-out and consent: months and six figures before you trade. And it is all sunk into the venue. If the location stops working, the owner walks away and the fit-out stays behind.
One AI runs the whole store.
It is the store manager, the cashier, the security guard and the marketing team in a single system. It opens the doors, tracks every product in real time, flags theft, responds to falls and decides which promotions to run. When stock runs low, it places the order and dispatches a human assistant to restock exactly what it asks for.
Humans don't run this store. They work for it.
App, card or phone entry
Customers unlock the door with the app, a bank card or their phone number. Secure, sensor-controlled contactless doors.
Unpaid-item detection
The system registers an unpaid item in real time, notifies the customer and alerts the operator in parallel.
Fall detection
Recognises a person down and pushes a medical alert, so there is safety cover with no staff.
Inventory & ordering
The AI tracks every product in real time and places the order before shelves run empty.
Scan & pay
Self-checkout POS with card & BLIK. The unit reconciles itself.
Promotions, picked by AI
The AI decides which offers to run and when, and pushes them to the in-store screens.
AI-powered vision that reads movement.
138-point skeletal tracking and behaviour recognition, built in-house over four years and running on commodity cameras at the edge.
An unpaid item, resolved in one tap.
The moment the system registers an unpaid item, the customer gets an instant notification: the item, the price, and a tap to pay. Security is alerted in parallel. Neutral by design, with no accusation, just the record and a one-tap fix. If it is wrong, one tap disputes it and a person takes over.
- Behaviour recognition trained on item-handling, not identities
- Runs on commodity cameras at the edge, with no specialist hardware
- Customer notification and operator alert fire in parallel
Person down. Help on the way.
The same skeletal model recognises a collapse and pushes an immediate medical alert. That is safety cover in an unmanned store, without a single camera ever identifying a face.
- Identity-free by architecture, and built to GDPR and EU AI Act requirements
- Critical for hospital, campus and late-night sites
- One model, many jobs: security, safety, inventory and fleet
Your brand. Any site.
Dropped anywhere.
Pharmacy
Swimwear & sun care
Clothing & merch
24/7 convenience
Books & stationery
Convenience, pharmacy, campus, forecourt or resort. The unit is the same, and the brand is yours.
Assortment configured per site and applicable law.
A partner stack that would take years to assemble.
NVIDIA
Good to know.
The things operators and brands ask us first.
How fast can a unit open?
Do you store faces or use facial recognition?
What hardware does it need?
Can it run under our own brand?
How many people does it take to operate?
What does a site need?
Can we license just the vision, without a container?
Can a unit be moved later?
Put your brand inside a store that runs itself.
Tell us the site and the brand. We'll show you a unit trading in weeks, not quarters.