Success story.
Cameras that understand what they see: 35% faster investigations and live alerts
A security company had hundreds of CCTV cameras and a clear idea: to stop just recording and start understanding. Forensic analysis, people counting, and real-time alerts, all powered by AI.
We take on any challenge
At Zenta there are no obstacles we cannot overcome.
Challenge
Finding an event in historical recordings consumed hours of manual work, accurately counting people required dedicated teams, and reacting in time required detecting incidents as they occurred.
Solution
We configured the Apple Fast VLM models so the cameras can interpret what they record. We built a forensic analysis pipeline with semantic search: simply describe the event with words to find it. We also integrated the CCTV system to send live alerts via email and internal applications.
Results from our management
35% less time spent on forensic investigations, with faster incident closures.
No dedicated hardware for counting people: 70% less CAPEX (estimated savings of US$38,000).
29% less response time to logistical incidents and +11 p.p. of accuracy in risk classification.
100% traceability and proactive fleet monitoring with live alerts.