I just released a working system that automatically finds the most engaging Instagram Reels from selected accounts — perfect for marketers, content creators, and media teams who want to turn viral Reels into their own hits.
How to Setup:
What this system does:
- Tracks Reels from a list of Instagram accounts
- Calculates performance vs. account median (views, likes, comments)
- Identifies early trends based on fast engagement
- Uses AI to:
- Classify each Reel (type + category)
- Transcribe & translate content
- Extract hooks (e.g., how attention is captured)
- Lets you confirm or reject ideas — approved ones go straight to your production table
How it works (simplified steps):
- I add Instagram handles to a Notion table — the automation fetches stats and recent Reels
- Each Reel is saved in a separate table with:
- Metadata (views, likes, comments)
- Engagement trend detection (early trend / trending / normal)
- AI-generated type and category (from a custom prompt)
- Auto-translated transcription + hook summary
- I review the results and click “Approve” on the best ideas — they move into my video production pipeline automatically
Customize it for your needs:
- You can change how “trending” is calculated — by default, it’s based on % above the account’s median stats
- AI classification uses a prompt — just update it in the Set Prompt node to match your own rules (e.g., business vs. entertainment)
What do you think — should I build a similar system for YouTube Shorts next?