Transcript-Based Summaries

Gistly reads a YouTube video's transcript and condenses it into a short summary using an underlying language model. A Beam agent requests a summary for a video named in the approved rule, for example a new upload from a channel the customer follows, and adds the result to a notes document or a notification for the person waiting on it. It works only from the transcript Gistly can access, not from the video's audio or visuals directly. Anything involving a video with no transcript available, or a summary that seems to miss the point, is routed to a human to watch and confirm.

Timestamped Key Points

Alongside a summary, Gistly can surface key points tied to specific timestamps in a video. A Beam agent takes these timestamped points and applies the approved rule, for example compiling them into a shareable outline for a training video or noting the timestamp where a specific topic comes up. It only presents what Gistly's summary returns, it does not judge which point matters most beyond the rule's criteria. Anything involving a request for deeper analysis of a specific moment, or a discrepancy between the summary and the video, is routed to a human to check directly.

Chapter-Level Breakdown

Longer videos can be split by Gistly into chapter-level summaries rather than one block of text. A Beam agent applies the approved rule to decide which chapters matter, for example the introduction and conclusion of a recorded webinar, and adds only those sections to a notes file or forwards them to whoever requested the recap. It draws solely on the chapter breakdown Gistly provides. Anything involving a chapter that seems mislabeled, or a video where the breakdown looks incomplete, is routed to a human to watch the relevant section before relying on it.