AI-Powered Daily Content Ideas with RAG and Webhook

AI-Powered Daily Content Ideas with RAG and Webhook

This workflow builds a Retrieval-Augmented Generation (RAG) system for generating and managing daily content ideas. It accepts new content via a webhook, chunks and embeds it into a Supabase vector store, and then allows an AI agent to retrieve relevant context and generate responses. The final output is logged to Google Sheets, and any errors are sent to a Slack channel. This is ideal for teams that need to curate, search, and augment a growing knowledge base of content ideas.

Node Breakdown

1. Sticky Note (n8n-nodes-base.stickyNote) — A visual note with placeholder text. No authentication required. It does not affect execution.
2. Webhook Trigger (n8n-nodes-base.webhook) — Listens for POST requests on /daily-content-ideas. No authentication (can be enabled manually). It triggers the workflow and passes incoming data (e.g., a new content idea text) to the next nodes.
3. Text Splitter (@n8n/n8n-nodes-langchain.textSplitterCharacterTextSplitter) — Splits the incoming content into chunks of 400 characters with an overlap of 40 characters. No authentication required.
4. Embeddings (@n8n/n8n-nodes-langchain.embeddingsCohere) — Converts each text chunk into a vector embedding using Cohere's embed-english-v3.0 model. Authentication: Cohere API key required via n8n credentials.
5. Supabase Insert (@n8n/n8n-nodes-langchain.vectorStoreSupabase) — Inserts the chunked and embedded data into a Supabase vector store index named daily_content_ideas. Authentication: Supabase credentials (URL and service role key).
6. Supabase Query (@n8n/n8n-nodes-langchain.vectorStoreSupabase) — Queries the same Supabase vector store to retrieve relevant chunks based on similarity. Authentication: Same Supabase credentials.
7. Vector Tool (@n8n/n8n-nodes-langchain.toolVectorStore) — Wraps the vector store query into a tool that an AI agent can call. No additional authentication.
8. Window Memory (@n8n/n8n-nodes-langchain.memoryBufferWindow) — Keeps a short-term memory of the conversation (buffer window). No authentication.
9. Chat Model (@n8n/n8n-nodes-langchain.lmChatOpenAi) — The large language model (e.g., GPT‑4) used by the agent. Authentication: OpenAI API key via n8n credentials.
10. RAG Agent (@n8n/n8n-nodes-langchain.agent) — Orchestrates the RAG process: uses the system message "You are an assistant for Daily Content Ideas", calls the Vector Tool to retrieve relevant context, and generates a response. Authentication: None (uses the connected Chat Model and tools).
11. Append Sheet (n8n-nodes-base.googleSheets) — Appends the agent's output to a Google Sheet (Sheet ID specified as SHEET_ID, tab Log). Authentication: Google OAuth2.
12. Slack Alert (n8n-nodes-base.slack) — Sends an error message to the #alerts channel if the RAG Agent fails. Authentication: Slack OAuth2 or token.

Setup Instructions

To use this workflow, you will need accounts and API keys for the following services:

• n8n — Self‑hosted or cloud instance.
• Cohere — Create an account at cohere.com and obtain an API key.
• Supabase — Create a project at supabase.com, enable the pgvector extension, and note your project URL and service role key.
• OpenAI — Create an API key at platform.openai.com.
• Google Sheets — A Google account and a spreadsheet where the SHEET_ID is replaced with your sheet's ID. The sheet must contain a tab named Log.
• Slack — A Slack workspace where you have permission to send messages to the #alerts channel. Generate a bot token or use OAuth.

1. Replace the placeholder SHEET_ID in the Google Sheets node with your actual sheet ID.
2. Configure each credential in n8n (Settings > Credentials) for Cohere, Supabase, OpenAI, Google Sheets, and Slack.
3. Set the webhook URL (generated by n8n) as the endpoint in your external system that posts new content ideas.
4. Adjust the chunk size and overlap in the Text Splitter as needed.
5. (Optional) Change the system message in the RAG Agent to match your use case.

Use Cases & Variations

• Content Curation — Use this workflow to ingest blog posts, social media captions, or product descriptions and then query them with natural language.
• Knowledge Base Chatbot — Replace the webhook with a chat trigger and memory to build a customer support Q&A bot.
• Idea Brainstorming — Store past brainstorming sessions or meeting notes, then ask the agent for new ideas based on that context.
• Error Handling — The Slack alert can be extended to log errors to a different channel or trigger another workflow for remediation.
• Alternative Embeddings — Swap Cohere for OpenAI embeddings (@n8n/n8n-nodes-langchain.embeddingsOpenAi) or other providers.
• Different Vector Stores — Replace Supabase with Pinecone, Qdrant, or Weaviate for production scaling.

{
  "name": "Daily Content Ideas",
  "nodes": [
    {
      "parameters": {
        "content": "Placeholder for Daily Content Ideas",
        "height": 530,
        "width": 1100,
        "color": 5
      },
      "id": "ca5f06fc-6dec-4477-bce5-e79c88a25e7d",
      "name": "Sticky Note",
      "type": "n8n-nodes-base.stickyNote",
      "typeVersion": 1,
      "position": [
        -480,
        -240
      ]
    },
    {
      "parameters": {
        "httpMethod": "POST",
        "path": "daily-content-ideas"
      },
      "id": "07884fea-1b21-4c07-9cc1-0f8c9546154d",
      "name": "Webhook Trigger",
      "type": "n8n-nodes-base.webhook",
      "typeVersion": 1,
      "position": [
        -300,
        0
      ]
    },
    {
      "parameters": {
        "chunkSize": 400,
        "chunkOverlap": 40
      },
      "id": "cddbaad6-33bf-4933-8a99-018c63325ad8",
      "name": "Text Splitter",
      "type": "@n8n/n8n-nodes-langchain.textSplitterCharacterTextSplitter",
      "typeVersion": 1,
      "position": [
        -130,
        0
      ]
    },
    {
      "parameters": {
        "model": "embed-english-v3.0",
// ... truncated (copy to see full JSON)