RAG-BASED CHATBOT WORKFLOW (n8n)

This n8n workflow implements a RAG (Retrieval-Augmented Generation) chatbot 🤖 that can answer questions based on your company documents or any dataset you provide. The bot uses Google Gemini embeddings, Pinecone vector store, and an AI Agent for contextual answers.

✨ FEATURES

• 📂 Upload and monitor company documents from Google Drive
• 🔄 Automatically process new or updated files
• 🧠 Convert documents into embeddings using Google Gemini API
• 💾 Store embeddings in Pinecone vector database
• 🗨️ Use AI Agent to answer questions based on retrieved documents
• 📝 Maintains conversation memory with sliding window buffer
• ⚡ Easy to extend to new datasets

🛠️ REQUIREMENTS

• n8n (latest version)
• Google Cloud Project with Vertex AI API enabled
• Google AI API Key (for Google Gemini)
• Pinecone account with index (e.g., company-files)
• Google Drive credentials in n8n (OAuth2)

⚡ SETUP STEPS

1. Google Cloud Project & Vertex AI API

   • Create a project in Google Cloud ☁️
   • Enable Vertex AI API ✅

2. Google AI API Key

   • Obtain your API key from Google AI Studio 🔑

3. Pinecone Setup

   • Create a free account at Pinecone 🖥️
   • Create an index called company-files 📌

4. Google Drive

   • Create a dedicated folder for company documents 📁
   • Note the folder ID 🆔

5. Configure Credentials in n8n

   • Google Drive OAuth2 🔐
   • Google Gemini API (PaLM) 🤖
   • Pinecone API Key 🗄️

6. Import Workflow

   • Import the provided JSON workflow into n8n 📥

7. Configure Nodes

   • Update Google Drive Trigger nodes to watch your specific folder 👀
   • Configure Pinecone Vector Store nodes with your index 💾

💡 USAGE

• Once the workflow is running, any new or updated document in the folder will be automatically processed.
• Ask questions through the chatbot 🗨️ and it will retrieve the most relevant information from your documents.
• If a question cannot be answered, the bot will respond:

  "I cannot find the answer in the available resources." ❌

📌 NOTES

• Ensure your Google AI API Key, Pinecone API Key, and Google Drive credentials are properly configured in n8n.
• Adjust the window buffer memory for longer or shorter conversation contexts.
• The workflow is fully modular, so you can add more tools, vector stores, or embeddings as needed.