How VideoQ AI answers work and what to check

For an uploaded video, VideoQ extracts the audio and transcribes it into timestamped text. The text is split into scenes and made searchable by meaning. A YouTube import retrieves existing subtitles rather than transcribing audio, and needs your own SearchAPI key.

When a question arrives, VideoQ follows these steps.

1. It checks that the person asking can use the course.
2. The AI searches scenes in the course and, as needed, reads surrounding subtitles or still images from uploaded videos.
3. It writes an answer from what it found, citing the source moments.
4. The student can play that part of the lecture from the cited time.

A citation does not guarantee that an answer is right. Audio can be misheard and scenes misinterpreted. Content shown on screen but not spoken, or brief actions, may also be missed. Before publishing, try the questions you expect from students, and compare the answers with the cited moments.

Only the course owner can read, export as CSV or delete saved questions and answers. Questions from invited members show who asked. Questions through a share link do not, even if the person is signed in, although the owner can still read whatever the question itself says.

Removing a member or disabling a share link does not delete history. To remove saved conversations, delete the course history separately. That applies to the whole course and does not erase CSV files already exported.

Answers use the course owner’s allowance. VideoQ reserves usage before writing an answer and attempts to return it if the answer cannot be completed. At the limit, new questions are refused from the owner, members and share link users alike. Heavy bursts can also be limited temporarily while allowance remains.