Video Navigator demonstrates a simple but new video watching experience. The main idea is to present a video side by side with an AI-generated overview that is fine-grained, interactive, and pure text. This interface supports watching, rewatching, skimming, getting a gist at a glance, extracting key points, and jumping between segments. Video Navigator is aimed at people trying to learn, like a student watching a lecture or a citizen watching the presidential debate.

Right now, this is a demo, not a standalone product. My hope is for platforms like YouTube to incorporate a similar video interface, accessible via a single click or toggle.

Besides improving the watching experience, fine-grained written overviews can help improve the video selection UX. Currently, a user is faced with a grid of videos represented by thumbnails and titles and they must quickly judge and select a video to watch. What if the user could hover over a video to see its chapters (AI-generated or not) and possibly its fine-grained details? A written overview is a key defense against attention hacking, as the current video selection UX makes it easy to fall prey to a catchy title/thumbnail and hard to quickly get a gist of a video's content.

Inspired by Bret Victor's dynamic poster comic strip of a video.