SENTINEL ingests live video from CCTV cameras above production machines and detects, in real time, whether each worker wears the equipment required for the task, flagging missing heat and cut resistant gloves and head coverings the moment a worker enters a monitored zone, and tracking hand position relative to hazardous machine zones. The AI approach is supervised object detection, with models tuned for the lighting, occlusion and motion of real shop floors. A second path fuses the RGB detections with thermal imagery reading oven and hot surface heat signatures, so the system reasons about burn risk by combining the position of a bare hand with nearby equipment temperatures. Detections become anonymous events and alerts that a supervisor acts on. Inference runs on premises and faces are blurred, so footage never leaves the facility. The product is at TRL 7, demonstrated at HELIOS, a bakery and pastry production facility, under real production conditions. The detection service runs on standard cameras and edge hardware with a containerised deployment, so onboarding a new plant is a matter of configuration.