n8n Visual Storytelling Content Factory with Gemini, Replicate AI, and Human-in-the-Loop Publishing

This n8n workflow automates the creation of visual storytelling content, leveraging AI models from Gemini and Replicate for content generation and image manipulation. It incorporates a human-in-the-loop approval step via Slack to ensure quality and relevance before final publication.

What it does

This workflow streamlines the content creation process through the following steps:

1. Manual or Scheduled Trigger: The workflow can be initiated manually or on a predefined schedule.
2. AI Content Generation:
   • It uses a "Basic LLM Chain" (likely powered by Gemini or a similar large language model) to generate initial content, potentially including text prompts or story ideas.
   • An "OpenRouter Chat Model" is then utilized, providing flexibility to interact with various AI models (including Replicate AI via OpenRouter) for further content refinement or image generation prompts.
3. Image Manipulation: The "Edit Image" node is used to process and modify images, which could involve resizing, cropping, adding effects, or compositing based on the AI-generated instructions.
4. Data Transformation:
   • "Edit Fields (Set)" allows for custom data manipulation and structuring of the generated content.
   • "Split Out" and "Loop Over Items (Split in Batches)" enable processing of multiple content pieces or image variations efficiently.
   • "Limit" and "Sort" nodes help in managing and organizing the generated items.
   • "Summarize" can be used to aggregate or condense information.
5. Human-in-the-Loop Approval:
   • The workflow sends a notification to Slack, presenting the generated content (text and images) for review.
   • An "If" node likely checks for approval or specific conditions before proceeding.
6. External Publishing (HTTP Request): Approved content is then published to a target platform via an HTTP Request, which could be a CMS, social media, or another content distribution service.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running instance of n8n.
• AI Service Credentials:
  • Access to Gemini AI (or a similar LLM) configured within the "Basic LLM Chain" node.
  • An OpenRouter API Key for the "OpenRouter Chat Model" to access various models, including potentially Replicate AI models.
• Slack Account: A Slack workspace and an n8n Slack credential configured for sending approval notifications.
• Target Publishing Platform: Credentials or API endpoints for the platform where the content will be published (configured in the "HTTP Request" node).

Setup/Usage

1. Import the Workflow:
   • Copy the provided JSON code.
   • In your n8n instance, click on "Workflows" in the left sidebar.
   • Click "New" and then "Import from JSON".
   • Paste the JSON code and click "Import".
2. Configure Credentials:
   • Locate the "Basic LLM Chain" node and configure your Gemini (or other LLM) credentials.
   • Locate the "OpenRouter Chat Model" node and configure your OpenRouter API key.
   • Locate the "Slack" node and select or create your Slack API credential.
   • Locate the "HTTP Request" node and configure the URL, headers, and body according to your target publishing platform's API documentation.
3. Customize Nodes:
   • Basic LLM Chain / OpenRouter Chat Model: Adjust the prompts and model parameters to generate the desired content and image descriptions.
   • Edit Image: Configure the image manipulation operations as needed (e.g., resizing, watermarking).
   • Edit Fields (Set): Customize the data transformation to match the input requirements of your publishing platform.
   • Slack: Customize the message content for the approval notification to include relevant details for human review.
   • If: Adjust the conditions for approval based on the Slack response.
   • Schedule Trigger: If using a schedule, configure the desired interval for content generation.
4. Activate the Workflow: Once configured, activate the workflow to start generating and publishing content. You can test it manually first using the "When clicking 'Execute workflow'" trigger.