Glossary
What is AI video summary?
Last updated
An AI video summary is a one-paragraph or bulleted recap of a video's content, generated automatically by a large language model reading the video's transcript. The summary appears at the top of the share page and lets a viewer decide in 10 seconds whether the video is worth watching, without sitting through 20 minutes of footage.
AI summaries work because the model reads the transcript, not the video. A 30-minute meeting produces a 2,500-word transcript; a one-shot prompt ('Summarize the key decisions and action items') produces a 200-word summary that captures the actual value of the meeting. The same transcript feeds the AI chapter markers (topic transitions, with timestamps) and the ask-the-video chat (a viewer types a question, the model answers from the transcript with a clickable timestamp). Together these are the three AI features that turn a passive video into a searchable, answerable artifact. The brief said 'every Pullsy recording ships with an instant transcript, an AI summary, ask-the-video chat, and a Markdown copy of the whole thing.' Related terms: video transcript, async video, video message.