See it in action.

Pick a video file. We push it to our streaming server, the inference engine runs on it live, and the annotated reflection plays back side-by-side. Stream is torn down when you close this page — nothing stored, nothing shared.