How to summarize a YouTube video with citations

A summary that cannot point back to where a claim came from is not much more trustworthy than a guess. Fetching the transcript first, with its timestamps kept, gives the summary something to anchor to: a chapter, a quote or a key point can link straight back to the second it was said.

AI Summary reads the transcript already in your history and returns an overview, key points, timestamped chapters, action items, quotes and open questions. Each chapter deep-links to the point in the video it covers.

Key points and action items are written as short, standalone statements, so they can be scanned quickly before deciding whether to watch the underlying chapter.

Ask This Source answers a specific question about the video and cites the timestamp the answer came from, so a claim can be checked in one click instead of scrubbing through the recording.

This is particularly useful for a long or technical video, where the answer to one question might otherwise mean scrubbing back and forth through several minutes of footage.

Standard quality runs on Claude Sonnet for 10 units; Deep runs on the largest available Claude model for 30 units and is worth the extra cost for a denser or more technical video. A cached repeat of the same source and quality costs 1 unit.

Both qualities return the same structure, an overview, key points, chapters, action items, quotes and questions, so switching between them changes how carefully the video was read, not what shape the result takes.

AI features are part of Pro or API; the free plan covers the transcript fetch itself. The unit cost of a summary is shown before it runs and reported in the response headers afterward.

Because the cost is shown up front, a Standard summary and a Deep summary of the same video can be compared on price before choosing one.