How does AI help with retail loss prevention?

Instead of reviewing footage after a loss is discovered, you ask questions across every camera and every day at once — who was behind the counter, which refunds happened without a customer present, when a stockroom door was opened — and you get answers linked to the exact moment on video.

Can it alert me while it is happening?

Yes. You write the rule as a plain sentence, such as 'alert me if anyone who isn't staff goes behind the counter after we close', and it messages you the moment it happens with the clip, the time and what it saw.

Does it work across several shops?

Yes. Each site runs its own machine with unlimited cameras on it, and each licence is a one-time $1,200 rather than a per-camera subscription.

Use case · Retail

Loss prevention without scrubbing footage

Most shrink is found late, if at all, because reviewing video costs hours nobody has. Asking a question costs seconds.

The three investigations that pay for themselves

After-hours access. Ask who entered, who went behind the counter, and whether they were staff — across a whole month, in one question.
Refunds and voids. Find the transactions where no customer was present at the till, then open the second of video that proves it.
Stockroom and delivery gaps. Compare what arrived on camera with what was booked in, including how many people were on the delivery.

Alerts you write in a sentence

Rules are plain English, not a rules engine: "alert me if the back door opens after 9pm", "tell me if the till drawer is open with nobody at the counter". Each alert arrives with the clip, the time and a description of what it saw, so you can judge it from your phone.

Evidence you can hand to someone else

Every statement in an answer carries a numbered citation that opens the moment on video, and anything inferred rather than observed is labelled. That is the difference between a hunch and something you can put in front of an insurer, a franchisor or an employee.

What it costs a shop

$1,200 once, per machine, with unlimited cameras on that machine and no per-question metering. A single site runs on a Mac mini M4 with 24 GB; larger sites use the recommended RTX 5060 Ti 16 GB build and handle over 100 cameras.

Frequently asked

How does AI help with retail loss prevention?: Instead of reviewing footage after a loss is discovered, you ask questions across every camera and every day at once — who was behind the counter, which refunds happened without a customer present, when a stockroom door was opened — and you get answers linked to the exact moment on video.
Can it alert me while it is happening?: Yes. You write the rule as a plain sentence, such as 'alert me if anyone who isn't staff goes behind the counter after we close', and it messages you the moment it happens with the clip, the time and what it saw.
Does it work across several shops?: Yes. Each site runs its own machine with unlimited cameras on it, and each licence is a one-time $1,200 rather than a per-camera subscription.

Keep reading

AI security camera search
The mechanics behind the questions.
On-premise vs cloud
Why per-camera subscriptions hurt multi-site retail.