CILAB launches XAIVA Agent for natural-language CCTV search
The agent runs inside CILAB's XAIVA platform on customer on-premise servers, letting operators query surveillance events in plain sentences and export PDF reports.
CILAB released XAIVA Agent, an AI agent for its vision AI-based video analytics platform XAIVA, the company said on November 5, 2026. The agent lets operators search surveillance events with natural-language sentences instead of manually reviewing CCTV feeds and alert logs.
The agent explains detected events in sentences and summarizes statistics and week-over-week changes. Results can be exported as a one-click PDF report. Users can also register internal safety rules and manuals so the agent answers related questions with supporting evidence.
A "worth checking now" section on the first screen highlights events that rose or fell compared with the prior week, repeated occurrences at the same location, and events clustered in time.
The XAIVA engine monitors site CCTV around the clock with more than 130 vision models that detect missing helmets, fire, fallen workers and entry into hazardous zones. XAIVA Agent runs within XAIVA without separate development or integration with an external service.
All analysis runs on customer internal servers, supporting closed-network on-premise environments so video and results do not leave the site. According to datanet.co.kr, XAIVA holds KISA intelligent CCTV performance certification and GS certification grade 1.
Quick answers
What does XAIVA Agent do?
It lets CCTV operators search surveillance events using natural-language sentences, explains detected events, summarizes statistics and week-over-week changes, and exports results as a one-click PDF report.
Does XAIVA Agent send video data outside the site?
No. All analysis runs on customer internal servers in closed-network on-premise environments, so video and results do not leave the site.