08-14-2026, 02:09 PM
The 4-step pipeline that works:
1. Get the transcript – YouTube Data API (captions) or yt-dlp
2. Transcribe if no captions exist – Whisper (openai-whisper) runs locally
3. Summarize with an LLM – chunk the transcript, ask for structured notes: key points, timestamps, action items
4. Store – Notion / Google Docs / Markdown files per video via API
You can wire the whole thing in n8n or Make: trigger on new upload → grab captions → summarize → post to your notes. Fully hands-off once it runs.
What's your video → notes setup? And does anyone have a clean trick for keeping timestamps in the notes?
1. Get the transcript – YouTube Data API (captions) or yt-dlp
2. Transcribe if no captions exist – Whisper (openai-whisper) runs locally
3. Summarize with an LLM – chunk the transcript, ask for structured notes: key points, timestamps, action items
4. Store – Notion / Google Docs / Markdown files per video via API
You can wire the whole thing in n8n or Make: trigger on new upload → grab captions → summarize → post to your notes. Fully hands-off once it runs.
What's your video → notes setup? And does anyone have a clean trick for keeping timestamps in the notes?