Automating AI-Generated Ideas & Post to Blog/Social with n8n – Part 2

Generating Content from Ideas

This section focuses on taking AI-generated ideas and converting them into full content automatically. The workflow includes:

1. Retrieving AI-Generated Ideas – Fetching stored ideas from Supabase.
2. Processing with an AI Model – Using LLM (e.g., OpenAI, Gemini) to generate structured content.
3. Storing Processed Content – Saving the generated content back to the database.
4. Sending Notifications (Optional) – Notifying users via Telegram about new content.

---

Step-by-Step Implementation

1. Retrieve AI-Generated Ideas

• Add Edit Fields and Split Out node if you process multi ideas or sub-topics.
• Add a Postgres node to fetch stored ideas from Supabase.
• Configure it to pull only unprocessed ideas.
• Add Telegram (Send a text message) node to receive log/error (optional).

2. Process Content with AI

• Use an Basic LLM Chain chain, connect with an AI model, e.g., Claude/DeepSeek/OpenAI/Gemini.
• Attach Structured Output Parser to format the AI-generated response.
• Define prompt instructing AI to expand ideas and generate structured content.

3. Parse AI Output

• Attach a Structured Output Parser node to format the AI-generated response.
• Ensure the output is organized into fields and ready to be saved in database.

4. Store Processed Content and Notify via Telegram

• Use a Supabase (Create a row) node to save the AI-generated content in Supabase.

• Send a Telegram notification with a preview of the generated content.
• Due to Telegram’s message sending limitations, you can add a Function node (code below) to process the content before the message reaches the Telegram node.

const maxLength = 4096; // Telegram's max message length

// Get your message text
const messageText = $('Previous node').first().json.content || items[0].json.yourMessageField;

// Check if it exceeds the max length and truncate if needed
const truncatedMessage = messageText.length > maxLength 
  ? messageText.substring(0, maxLength - 3) + "..." 
  : messageText;

// Return the truncated message
return {
  json: {
    ...items[0].json,
    truncatedMessage: truncatedMessage
  }
};

---

Error Handling & Optimization

• If the AI model fails, add a Telegram (Send a text message) node to send log/error.
• Set up retry mechanisms for node failures.

---

Conclusion

This workflow streamlines AI-generated content creation, storing processed text for later use in blogs, social media.