Burn subtitles into your videos without manual editing

The agent calls the ZapCap API with the key and sees three things there: subtitle style templates, uploaded videos and task statuses — transcription, render and clip jobs. It takes the source from a public link and hands it to the service, and picks up the finished MP4 from the link in the response once the render completes. A batch of videos goes through this whole path with one chosen template for the entire pack.

ZapCap works as a pipeline, and that is the main thing the agent accounts for: first the upload, then the transcription, then a separate step approving the transcript, and only after that the render. An accepted task is not yet a finished file, so the agent waits for the status instead of treating the service response as confirmation of a result. The source language is detected during transcription, and a version with subtitles in another language is built from it, without rebuilding the video by hand.

On the writing side the agent does everything the pipeline needs: it uploads videos, creates subtitle and clip tasks and approves the transcript before rendering. Clipping is capped by the service itself — up to five short clips from a single video within the chosen duration range. The agent does not design style templates, it picks from the ready ones and applies the same one across the batch so the videos do not drift apart in style.

Rendering is paid for separately from the subscription and by the minute of finished video, so the agent estimates the cost of a batch upfront — from the duration of the sources rather than their count. A task is accepted into the queue instantly, but the file appears minutes later, and on long videos the agent polls the status instead of holding the wait inside one request. If credits ran out and a render was rejected, it names the reason directly instead of showing an empty task list.