Automated Content Repurposing for YouTube → LinkedIn

Problem

A solo creator with 40k YouTube subscribers was publishing one video per week but almost nothing on LinkedIn. He knew he was leaving reach on the table — the audience is there, but repurposing felt like a second job.

The constraint: he'd tried hiring an editor. The output was inconsistent, slow, and expensive. He wanted something he could trust to run without supervision.

Solution

A fully automated content repurposing pipeline triggered by a new video upload:

1. YouTube trigger — n8n watches the channel; new video fires the workflow
2. Transcription — Audio extracted via yt-dlp, transcribed with Whisper API
3. Content generation — Five separate GPT-4o prompts, each with a distinct persona and format:
   • Long-form LinkedIn post (insight-led, 800 words)
   • LinkedIn carousel outline (5 slides with hooks)
   • Twitter/X thread (10 tweets)
   • Email newsletter section (250 words)
   • Short-form script for a YouTube Short
4. Review queue — All outputs land in a Notion database with status "Draft". Creator reviews, approves, tweaks.
5. Scheduled publish — Approved posts go to a Buffer queue for optimal timing

Result

8× weekly content output (1 video → 5+ pieces) Zero additional time from the creator Consistent voice — no brand drift between pieces

After 6 weeks: LinkedIn following grew 2.3× and newsletter open rate went from 28% → 41% (more relevant content hits the right audience at the right time).

What made it work

The "review queue" pattern is critical. Full automation without human review produces slightly-off content that erodes trust. The sweet spot: AI generates 90% of the work, human approves in 10 minutes.

Prompt engineering was 60% of the project. Each format needed its own persona, constraints, and output structure.

Stack

• n8n — workflow orchestration
• Whisper API — transcription
• GPT-4o — content generation (5 prompts)
• Notion — review and approval queue
• Buffer — scheduled publishing
• YouTube Data API — video metadata