Loss prevention a second early.
For the small grocers, pharmacies, and corner stores who deserve better than reviewing footage after the loss. AIsle reads the behaviors that lead to theft — and tells your floor staff in under two seconds.
<2s
$0
100%

$9.1B
Canadian retail loss to theft, 2024
+66%
Police-reported shoplifting, 10-year trend
95%
of retailers cite ORC as their #1 threat
Theft is up across Canada.
Cameras only watch the aftermath.
Canadian retailers lost $9.1 billion to theft in 2024 — up from $5B six years ago. Police-reported shoplifting hit 182,361 incidents, up 14% in a single year. Your cameras are recording it. They aren't preventing it.
CCTV is reactive.
Footage is reviewed after the loss. By the time you have a face, the merchandise is gone.
Tag-based EAS misses 40%+ of incidents.
EAS works for tagged SKUs. Most retail loss happens around concealment, sweeping, and short-cons that tags can't see.
Floor staff can't watch every aisle.
Theft windows are 20–30 seconds. A second pair of eyes — at machine speed — is the missing piece.
Capture. Detect. Alert.
A single pipeline from existing IP cameras to a notification on a floor manager's phone — most events surface in under two seconds.
Capture
AIsle taps into existing IP cameras over RTSP/ONVIF. No new hardware. No DVR replacement. Most stores live in an afternoon.
Detect
Open-source CV stack (RT-DETR · ByteTrack · VideoMAE — Apache 2.0 + MIT) flags concealment, sweeping, loitering, queue-skipping. Behavior, not faces.
Alert
A 4-second clip plus a one-line description lands on the floor manager's phone. Eighty percent of shoplifting events end the moment a staff member walks over to ask for help.
One screen for the floor team.
New events appear at the top. Filed events drop off. Your job is to walk over and ask if they need help finding something.
Store 04 — Kelowna West Grocery
6 cameras · 23 ms · today
- Today, 14:23CAM 03Concealment · Aisle 4 · 23 s loitering
- Today, 13:51CAM 05Sweep · repeated reach to high shelf
- Today, 13:11CAM 02Loitering · resolved by staff touch
- Today, 12:44CAM 03Cart abandonment · recovered
- Today, 11:03CAM 01Anomalous queue dwell
Behavior, not faces.
Facial recognition has been ruled against by Canada's Privacy Commissioner. AIsle doesn't ship it — and it isn't a missing feature, it's the design. AIsle is fully compliant with PIPEDA and BC PIPA, out of the box.
We don't recognize faces.
AIsle is trained on actions and movement, not identities. There is no face database. There can't be — we never extract one.
Your video doesn't leave the store.
On-prem inference is the default for Pro and Enterprise. The cloud sees alerts, not raw footage. Data residency stays in Canada.
We tell you when, not who.
An alert says "Aisle 4, 23 seconds of loitering" — not "that person." Your staff decides who to approach.
Built for the categories where it hurts most.
Grocery
Independent · 6–24 cameras · live in an afternoon
ReadConvenience
High violence per incident · alerts before sweeps
ReadLiquor + cannabis
Regulated · data residency in-store via Edge
ReadPharmacy
Highest per-incident loss · OTC, formula
ReadHardware
Tools, copper, batteries · wide aisles, blind spots
ReadThe numbers from our pilot stores.
−63%
shrink reduction in 12-week pilot · n=4 stores · Q1 2026
<2s
median alert latency from event to staff phone
$0
in new hardware required to install AIsle Cloud
Per camera, per month. No hardware tax.
Pro
C$99
Unlimited cameras. Priority alert queue. On-prem option. Multi-store dashboard.
Start a 30-day pilotEnterprise
Custom
Multi-region rollouts, SLA, dedicated success manager, on-prem mandatory.
Talk to salesCompare full plans on the pricing page.
Catch it before it happens.
30-day pilot in your store. No new hardware. No long contract. Live in an afternoon, results in a fortnight.