Built a Human-in-the-Loop AI Publishing Agent in n8n that combines AI-powered content creation with human review and approval before publishing.
The workflow receives content requests through Telegram and uses a dedicated AI agent to research and generate social media content. Instead of publishing AI-generated content immediately, the system introduces a human approval stage where the user can review the generated post and decide whether to approve it or request changes.
A text classifier interprets the feedback and routes the workflow accordingly. Approved content is automatically published to X, while rejected or revised content is sent to a dedicated Revision Agent, which uses the feedback to improve the post before returning it to the review cycle.
Key capabilities:
• AI-powered social media content generation
• Web research for context-aware content
• Human review and approval workflow
• Telegram-based feedback collection
• AI classification of approval decisions
• Automatic content revision based on feedback
• Iterative review and revision loop
• Automated publishing to X
• Multi-model AI architecture
• Modular n8n workflow design
The result is a controlled AI publishing system that combines the speed of automation with human oversight, helping ensure content is reviewed and refined before it goes live.
Like this project
Posted Sep 27, 2026
Built a Human-in-the-Loop AI Publishing Agent in n8n that combines AI-powered content creation with human review and approval before publishing.