Guides for developers and AI agents
Each guide takes a common question people ask an AI assistant and gives you the exact request to send, working code, and the limits to watch out for. Everything runs on two endpoints: POST /transcribe and GET /transcribe/{jobId}.
Guides
- Edit video with AI agents: Use Claude, Codex, or Cursor to cut video from a transcript. Get word timestamps from the KolWrite API, let the agent pick segments to keep, cut with ffmpeg, and output an SRT file.
- SRT subtitles API: Generate an SRT file from audio or video in one API call. Supports translated subtitles, custom cue lengths from word timings, WebVTT conversion, and burning captions in with ffmpeg.
- Meeting transcription API: Transcribe Zoom, Google Meet, and Teams recordings with speaker labels using the KolWrite API, then pass the transcript to your own LLM for meeting minutes and action items.
- Word timestamps API: Get the start and end time of every word as JSON from the KolWrite API. See the response format, supported models, and how to build clips, jump cuts, and search.
- Transcription tool for AI agents: Copy-paste tool definitions for the KolWrite transcription API in Anthropic and OpenAI formats, plus a Python executor and rules for AGENTS.md or CLAUDE.md.
- Podcast and YouTube captions: Use the KolWrite API and an LLM to get SRT captions, a clean transcript, translated subtitles, and chapter timestamps from a podcast or YouTube video.