Using custom Amazon AI models trained on the video segments, subtitles and dialogue of the Prime Video library, it can create descriptions of key events, places, times and even conversations. It then applies guardrails to ensure it is spoiler-free and easy to read at a distance.