I’ve been using n8n for content automation and finally got this working cleanly enough to share.
What it does:
Two workflows, one Telegram bot:
-
YouTube Summary Bot — Send a YouTube URL to your Telegram bot. The workflow grabs the transcript (using the YouTube Transcript node), sends it to Claude Haiku, and returns a 5–7 bullet summary directly to Telegram. I use this while commuting — send the video, read the summary when I arrive.
-
LinkedIn Post Generator — Send
/linkedin https://article-urlto the same bot. The workflow fetches the article, strips the HTML, and drafts a 250-word LinkedIn post in engineering voice (hook + 3 learnings + closing question). No hashtags by default — I add those manually.
Both run on the same Telegram bot token and same Claude API key. One import, one credential config.
Nodes used:
- Telegram Trigger + Telegram
- HTTP Request (for article fetching)
- Code node (for transcript parsing)
- Claude / Anthropic Chat Model
- IF node for routing the two flows
Setup time: ~10 minutes if you already have an n8n instance.
Why I built this: I write about AI/data engineering on LinkedIn. The 20-minute gap between “I watched something useful” and “I wrote a post about it” was killing my consistency. This cuts it to 30 seconds.
Sharing the workflow JSONs for free — just drop a comment if you want them and I’ll DM. Also wrote PDF setup guides if the JSON alone isn’t clear enough.
Happy to answer questions about the implementation — the article fetching part required some cleanup to handle paywalls gracefully.
