Automated X Creator / Human-in-the-Loop Slack

Automated X Content Creation with AI Agents & Human Approval This workflow automates the entire lifecycle of creating and publishing content for an X (formerly Twitter) personal brand. It functions as a sophisticated content pipeline, beginning with AI-driven idea generation and culminating in automated posting and data logging. It uses a multi-agent AI system for drafting and refinement, incorporates a human-in-the-loop approval process via Slack, and uses Google Sheets as a database for content history and stylistic examples, creating a self-improving system over time. Technologies Used - AI & Language Models: OpenAI (GPT-4.1-mini, GPT-4o-mini), n8n LangChain Nodes (Agents, Chains, Memory) - Collaboration & Notification: Slack - Social Media: X (Twitter) - Data Storage: Google Sheets, Notion - Automation Core: n8n (including sub-workflows) How AI is Integrated This workflow utilizes a powerful multi-agent AI architecture where different AI models and agents have specialized roles, collaborating to produce high-quality, on-brand content. - AI Idea Creator: An initial AI chain (LLMChain) reads a brand brief (fetched from Notion) and generates a list of 10 relevant and engaging post ideas. It uses a Structured Output Parser to ensure the ideas are delivered in a clean, usable JSON format. - AI Content Creator (Agent 1): This is the primary drafting agent. It takes a single post idea and is governed by a complex set of instructions. Its key AI-driven functions are: - Autonomous Tool Use: It can independently decide to use tools like Get_Brand_Brief to understand the brand voice and Check History (Google Sheets) to avoid posting duplicate content. - AI-Powered Quality Scoring: The agent uses the Get_Content_Feedback tool, which calls a separate sub-workflow. This sub-workflow employs another AI model to analyze the drafted post against the brand brief and return a quality score (0-1) and actionable feedback. - Iterative Refinement: Based on its instructions, the agent will loop, using the AI-generated feedback to rewrite the post until the quality score meets the required threshold (e.g., > 0.7).