Personal Portfolio / CV RAG Chatbot with Conversation Store and Email Summary

This n8n workflow creates a sophisticated RAG (Retrieval Augmented Generation) chatbot for a personal portfolio or CV. It allows users to interact with a chatbot that can answer questions based on documents stored in Google Drive, maintain conversation history in NocoDB, and send email summaries of conversations via Gmail.

What it does

This workflow orchestrates several key functionalities to provide an intelligent and interactive chatbot experience:

1. Receives User Queries: It acts as a webhook endpoint, waiting for incoming HTTP requests that contain user questions.
2. Manages Conversation History: Utilizes NocoDB to store and retrieve past conversation turns, providing context to the AI agent.
3. Retrieves Information from Google Drive: It can be configured to load documents from Google Drive on a scheduled basis. These documents are then processed to create a knowledge base.
4. Processes Documents for RAG:
   • Loads documents using a Default Data Loader.
   • Splits the document content into smaller, manageable chunks using a Recursive Character Text Splitter.
   • Generates embeddings for these text chunks using Embeddings Google Gemini.
   • Stores these embeddings in a Pinecone Vector Store to enable efficient semantic search.
5. Answers Questions with AI Agent:
   • Employs an AI Agent powered by a Google Gemini Chat Model.
   • The agent uses a Vector Store Question Answer Tool to retrieve relevant information from the Pinecone knowledge base based on the user's query and conversation history.
6. Responds to User: Sends the AI-generated answer back to the original webhook caller.
7. Summarizes Conversations (Optional/Manual Trigger): A separate branch of the workflow, triggered by a Schedule Trigger, can summarize conversations and send them via Gmail. This part of the workflow is currently disconnected from the main chatbot flow based on the provided JSON, implying a manual or separate trigger for summaries.

Prerequisites/Requirements

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

• n8n: A running n8n instance.
• Google Drive: To store your portfolio/CV documents.
  • Requires a Google OAuth2 credential configured in n8n.
• NocoDB: To store conversation history.
  • Requires a NocoDB API Token credential configured in n8n.
• Pinecone: To store and retrieve document embeddings for RAG.
  • Requires a Pinecone API Key and Environment configured in n8n.
• Google Gemini (or compatible LLM): For generating embeddings and powering the AI agent.
  • Requires a Google Gemini API Key credential configured in n8n.
• Gmail: For sending conversation summaries.
  • Requires a Google OAuth2 credential configured in n8n.

Setup/Usage

1. Import the Workflow: Import the provided JSON into your n8n instance.
2. Configure Credentials:
   • For each node requiring credentials (Google Drive, NocoDB, Pinecone, Google Gemini, Gmail), click on the node and select or create the appropriate credential.
   • Ensure your Google OAuth2 credentials have access to both Google Drive and Gmail.
3. Google Drive Setup:
   • Specify the folder ID in the Google Drive Trigger node where your portfolio/CV documents are located.
   • The Google Drive node should be configured to read the content of these documents.
4. NocoDB Setup:
   • Configure the NocoDB node to connect to your NocoDB instance and specify the table where conversation history will be stored.
5. Pinecone Setup:
   • Configure the Pinecone Vector Store node with your Pinecone API Key, Environment, and the index name you wish to use.
6. Google Gemini Setup:
   • Ensure your Embeddings Google Gemini and Google Gemini Chat Model nodes are configured with your Google Gemini API Key.
7. Activate the Workflow:
   • Activate the Webhook node to start listening for incoming chat queries.
   • Activate the Schedule Trigger if you want to enable automatic conversation summaries.
8. Trigger Document Ingestion: Manually execute the branch starting with Google Drive Trigger to ingest your documents into the Pinecone vector store initially. This should also be scheduled for periodic updates if your documents change.
9. Interact with the Chatbot: Send HTTP POST requests to the webhook URL provided by the Webhook node with your user questions. The chatbot will respond based on your documents and conversation history.
10. Review Email Summaries: If the scheduled summary is active, you will receive email summaries of conversations in your configured Gmail account.

Note: The "Sticky Notes" in the workflow JSON are for documentation purposes within n8n and do not require configuration.