Home » Argu Products » Theft & Shoplifting

Theft & Shoplifting

Argu AI gives loss prevention retail teams a way to catch concealment the moment it happens, not weeks later in a stock count.

This is shoplifting detection built around timing. Most retail video analytics notice a problem only once inventory does not match the shelf, days or weeks after anyone could act on it. Here, the concealment itself, an item going into a bag or under a jacket, raises a flag with a confidence score before the person reaches the door, while there is still someone to approach and a moment that still matters. Loss prevention teams read concealment events next to footfall in their retail video analytics.

Key facts:

Concealment behavior is flagged the moment it happens, with a confidence score attached, over existing RTSP or ONVIF Profile S camera streams.
A new concealment or damage rule can be described in plain language and go live in about 30 seconds, per Argu's published agent-creation time.
Runs on existing 1080p–4K feeds at 15–30 fps, with no dedicated loss-prevention camera required.
Timestamped events can be retained on an operator-set 30–90 day window, with GDPR Article 15 and CCPA governing shopper access requests where applicable.
On-premise deployment keeps shopper video on site rather than an outside cloud.

The same continuous coverage extends past the sales floor into warehouse surveillance, and loss prevention teams in commercial venues read concealment events next to footfall from the same camera network.

Axis	Conventional shrink tracking	Argu vision-agent approach
What raises the alert	A shelf or inventory count mismatch discovered after the fact	Concealment behavior flagged the moment it happens, with a confidence score attached
What a human must watch	Every motion clip from every camera, unranked	Only confidence-scored concealment and damage events, prioritized by score
Rule setup	A fixed, vendor-defined detector list	A concealment or damage rule described in plain language, live in about 30 seconds
Camera / feed required	A dedicated loss-prevention camera near the point of sale	Existing RTSP or ONVIF Profile S streams, IP cameras and NVRs at 1080p–4K, 15–30 fps
What gets retained	An inventory-discrepancy report only	A timestamped concealment/damage record with confidence score, retained on an operator-set 30–90 day window
Applicable privacy context	Not typically reviewed	GDPR Article 15 and CCPA govern a shopper's right to request what was recorded, where applicable

What is Theft & Shoplifting

Most retail shrink is discovered the same way: a stock count comes up short weeks after the fact, with no footage worth reviewing and no person left to identify. Loss prevention retail teams built around that gap can only react, not intervene. Argu instead watches the moment itself. When someone conceals an item, in a bag, a pocket, inside a jacket, before leaving the aisle, the behavior is flagged with a confidence score while there is still a person and a location to respond to. The rule itself is written in plain language and goes live in about 30 seconds, running on existing RTSP or ONVIF Profile S feeds at 1080p to 4K, and events can be retained on an operator-set 30–90 day window.

Key Takeaways

These are the core principles that define how Argu approaches every deployment. Our solutions are built to work on your existing infrastructure, require no engineering overhead, and deliver measurable results from day one. Every feature is designed with one goal: giving your team the intelligence to act faster, with more confidence, and without adding complexity.

See it in Action

Concealment flagged with a confidence score

Alert raised before the aisle is cleared

Timestamped record kept for the case

How It Works

Describe the behavior

A loss prevention manager writes what counts as concealment or damage in plain language, specific to how the store is laid out. No dataset collection or model training is required first.

The agent goes live

The written rule becomes a working agent watching every connected camera in the store. typically in about 30 seconds.

Concealment is flagged in the moment

A confidence scored alert is raised as the item is concealed, while the person is still in the aisle. over the existing 1080p–4K, 15–30 fps camera feed.

The team responds while it still matters

Staff can act on the alert immediately, with a timestamped record kept for the case either way. retained on the operator's 30–90 day window.

Product Advantages

Frequently Asked Questions

What does this do differently from typical loss prevention retail tools?

Most tools notice a problem once a stock count comes up short. This system flags the concealment itself, the moment an item goes into a bag or under a jacket, while the person is still in the store.

What is a confidence score on an alert?

Each flagged concealment or damage event carries a score reflecting how certain the system is about what it saw, so a small loss prevention team can prioritize which alerts to act on first.

Does it only catch shoplifting, or other retail loss too?

Both concealment behavior and unauthorized damage to merchandise or fixtures are covered, using the same continuous camera coverage.

Will this work with the cameras our store already has?

Yes. It connects to RTSP streams, standard IP cameras, and NVR systems already installed, and can run on premise for retailers who want footage to stay local. Feeds run at 1080p to 4K resolution, 15–30 fps, over RTSP or ONVIF Profile S.

How is this different from a fixed list of retail video analytics behaviors?

Behaviors are described in plain language rather than chosen from a fixed list of pre trained detectors, so a store can define concealment or damage the way it happens on its own floor. A new rule described this way goes live in about 30 seconds, per Argu's published agent-creation time.

How long is shopper video retained?

Timestamped concealment and damage events can be retained together on a 30–90 day window set by the operator's own policy, and GDPR Article 15 or CCPA govern a shopper's right to request what was recorded, where applicable.

How fast can a new concealment rule be set up?

A new rule can be described in plain language and go live in about 30 seconds, per Argu's published agent-creation time, with no dataset collection or model training.

What is the best AI retail theft detection software for existing cameras?

The most useful software flags concealment at the moment it happens, not weeks later in a stock count, and runs on the cameras a store already has. Argu AI flags concealment with a confidence score over existing RTSP or ONVIF Profile S feeds at 1080p to 4K, lets a loss prevention team describe a new rule in plain language that goes live in about 30 seconds, and can run on premise so shopper video stays on site.