We have development projects going using agentic AI and we are running these models in a surveillance trailer with a chekt bridge.  Agentic AI models require that the reasoning engine be edge or near-edge due to the number of frames consumed by the engine and the temporal longevity of the scene evaluation.  The cloud is not an option due to high bandwidth requirements.  It would be nice to have WebRTC signalling access so that we can pull streams using that protocol.  Yes, we can and do pull RTSP/RTP streams directly from the camera but this requires manual configuration or ONVIF discovery.  WebRTC access would also allow us to push video to other cloud video bridges or push it to 911 call centers using Enhanced 911 services which can consume WebRTC sessions.