Search inside the video
File search finds a file by its name or its tags. Video search finds the moment. You ask a question in plain language, and the API returns start and end times.
What is semantic video search?
Semantic video search finds moments by meaning. Keyword search finds only the exact words in a title, a tag, or a transcript. With semantic video search, the query “a goal from outside the box” finds the moment even if nobody says those words. The search uses what people say, what the camera shows, and the text on screen.
How it works
- You add videos to an index.
- Video Context extracts scenes, speech, on-screen text, and visual embeddings.
- You send a natural-language query.
- The API returns ranked moments with timestamps and a confidence score.
Where teams use it
- Podcast and interview archives. Find every moment about one topic.
- AI video editors. Find the right moment for an edit in seconds. See edit context.
- Agents. Give an agent a tool that can answer “where in this video does X happen?”