Case Study PLEXOR

The cameras a store already owns, turned into a live profit engine.

PLEXOR connects to a retailer's existing cameras and POS, then reads live video and transactions together. Theft, staff performance and safety surface as alerts in seconds, in plain language. No new hardware, and it deploys in hours. Zain built the backend and the AI platform behind it.

Client
PLEXOR
Sector
Retail · Visual AI
Engagement
Backend & AI platform
Stage
Live
Platform
Cloud · existing cameras
The PLEXOR marketing hero: existing retail security cameras turned into a real-time visual AI layer for theft, performance and safety
One AI layer over the cameras a store already runs: live video and POS read together, in real time.
By the numbers
~3sAlert response
0New cameras needed
3Focus areas: theft, productivity, safety
HoursTo deploy

The cameras are already running. They just never act on what they see.

Retailers bleed profit in places a recording never catches. Under-ringing at the register, shoplifting on the floor, sweethearting between staff and friends. Disengaged employees. A spill or a loiterer that turns into an incident. Operations that drift out of step.

A traditional camera system records and waits. The footage exists, but nobody watches it until something has already gone wrong, and by then the money is gone. One manager cannot stare at every feed across a shift, let alone match a moment on screen to the line on a receipt.

So the losses stay invisible. They blend into shrink and bad days, and the store keeps paying for them without ever seeing the cause.

A Vision-Language AI that reads video and the register as one stream.

Zain built the backend and the AI platform behind PLEXOR. The system pulls live video and transaction data together, so an event on a camera and the matching line at the POS arrive as one piece of evidence. It interprets context, then explains what it sees in plain language, in seconds, on cameras the store already owns.

What that meant in practice

  • Theft alerts matched to the POS line and the footage that proves them, the moment they happen
  • Performance tracking that watches punctuality, greetings and engagement so managers can coach
  • Safety monitoring that flags spills, weapons and loitering before they escalate
  • Plain-language read-outs of each event, on existing IP cameras, with no new hardware to install
The PLEXOR product homepage showing live retail intelligence: theft, performance and safety events surfaced from store cameras
The product

Live retail intelligence, in one view.

Each event shows up where an owner can act on it: a theft alert tied to its transaction, a greeting that got skipped, a hazard on the floor. The figures shown in the product, like an $8.42 under-ring or a 15% drop in greeting rate, are example values that illustrate the read, not audited customer results.

Theft · productivity · safety in one feed
The PLEXOR system architecture diagram showing existing IP cameras and POS feeding a Vision-Language AI that produces real-time alerts
Architecture

Existing cameras in, real-time alerts out.

The platform reads existing IP camera feeds and POS data, runs inference on both, and pushes alerts in seconds. Nothing new gets bolted to the ceiling, so a store can connect what it has and start seeing events the same day.

Existing IP cameras + POS · real-time inference
See it work

Walkthrough: live video and POS data turned into real-time retail alerts

The hardest part

Tying a single frame of video to the exact POS line it belongs to, across cameras of different makes and stores of different layouts, then deciding in about three seconds whether the moment is an under-ring, a missed greeting or nothing at all. That alignment, fast and reliable enough to act on, is where the real work lives.

Ego Eimi Engineering note · the build

Live, in front of the retail industry.

PLEXOR is live and reads existing cameras and POS today, with no new hardware between a store and its first alert.

The company rebranded from SurveilX, raised backing from iTech Ventures, and filed for two patents. It showed at NRF in New York in 2026, and counts Jordan Bolch of RaceTrac among its advisors. The product carries the story: a store connects what it already owns and starts seeing theft, performance and safety events the same day.

Vision-Language AI Computer vision IP cameras POS integration Real-time inference Cloud
Our role
Backend & AI platform
Team
Lead + specialists
Model
Product build
Status
Live
Project partner Zain Raza

Considering something similar? Explore a fixed-price build, an embedded engineering team, or the guide to building software without a tech team.

Next / Your build

Taking on new builds

Have something in mind?

Tell us what you're making. We reply within a day with a fixed price and a date.