Keeps: your cameras start speaking
By the JABB team5 min read

From image to action in 18 seconds
Keeps connects to your existing CCTV. A vision-language model watches, understands, and alerts in French, Arabic, or English. No hardware to buy.
What Keeps sees
- Footfall & queues : wait time > 3 min, abandoned queues, hour-by-hour peaks.
- Hygiene & compliance : gloves, HACCP, temperature zones.
- Security & incidents : intrusions, falls, abandoned packages.
Built for trust
No facial recognition - behaviors and objects only. Declared to the CNDP, limited retention, local processing (local YOLO × VLM).