AI video surveillance and occupational safety management for industrial plants.

The project

GuardIA is a platform we built for an industrial-sector client in Colombia. It connects the plant cameras to artificial intelligence to detect risky situations and alert the safety team on the spot, and brings occupational health and safety (OHS) management for the contractors entering the plant into one place.

The IP cameras connect to a local device at the plant that detects people in restricted zones and outside permitted hours. Each camera has its own AI alerts, written in plain language —for example “person without a helmet”, “person without a high-visibility vest” or “person less than 3 meters from the forklift”— and the AI checks every event against that camera’s alerts. When it finds one, the OHS team gets a notification on their phone and the video of the event is recorded as evidence.

What we built

• Detection of people in restricted zones, by schedule.
• Per-camera artificial intelligence alerts, configurable in plain language (PPE, distance to equipment, falls, smoke).
• Mobile notifications and video recording of each event.
• Contractor OHS management: worker registration, expiration tracking, work method statements with digital signatures and site access authorization.
• Site induction training for visitors and contractors, with assessment and certificate.
• AI usage control with a spending cap split across cameras.
• Installable mobile app (PWA) with role-based access.

Technology

Next.js, React and TypeScript (installable web app); Django and Django REST Framework (backend); local Python device with computer vision (YOLO); Anthropic’s artificial intelligence (Claude) to analyze the images; push notifications.

Next.jsReactTypeScriptDjangoDjango REST FrameworkPythonYOLO (computer vision)Claude (Anthropic)Installable app (PWA)