Let your team focus on customers while our AI + remote reviewers handle loss prevention
ShrinkHalt combines edge‑based action detection with a dedicated remote review team. Your staff stays on the floor, not watching monitors — we flag potential theft events and verify them before any notification.
Edge AI detection
Our lightweight AI runs on low‑cost edge devices in your store. It analyzes live video from existing security cameras to identify high‑risk actions: concealment into bags or clothing, consumption, and other suspicious behavior.
Remote human validation
When the AI flags a potential incident, our remote loss prevention team reviews the actual footage. They verify whether the alert is genuine — and only then send an actionable notification to your store staff.
Uninterrupted customer service
Your associates receive verified alerts only for real incidents. No constant CCTV monitoring, no distraction from helping customers. ShrinkHalt works in the background, reducing loss while keeping your team focused on sales and service.
Built for real‑time retail
The system is designed to run efficiently on edge hardware, processing video locally. Alerts are reviewed by humans in near real‑time, so your team can respond quickly when it matters.
Skeleton‑based action recognition
Our technology uses pose estimation to track body movements — shoulders, wrists, hips — rather than analyzing raw video. This approach focuses on motion patterns that indicate concealment or theft, while being less sensitive to lighting, camera angle, or background clutter.
The system classifies short windows of movement into operational categories: Floor Activity (normal browsing), Consumption, Concealment in Bag, or On‑Body Concealment. The AI runs locally on edge devices, and flagged events are sent to our remote review team for final verification.
The AI acts as a first filter — it dramatically reduces the number of events that need human review, but every critical decision involves a trained remote reviewer. This hybrid design balances automation with accountability.