Really nice chat too to blend it all together in context. This is compelling. I wonder if you could do two party video input and infer a bit about each speaker in order to use it gather all the information of the video to represent both parties in chat.