Ask Questions About a PDF Using AI

This n8n workflow enables you to interact with PDF documents stored in Google Drive using AI. It allows you to upload a PDF, embed its content into a Pinecone vector store, and then ask questions about the PDF, receiving AI-generated answers based on the document's content.

What it does

This workflow automates the following steps:

1. Triggers on chat message: Initiates the workflow when a chat message is received, allowing for user interaction to either upload a PDF or ask a question.
2. Loads PDF from Google Drive: Retrieves a specified PDF file from Google Drive.
3. Processes PDF content: Converts the binary PDF data into a readable document format.
4. Splits document into chunks: Divides the document into smaller, manageable text chunks for efficient processing by the AI.
5. Generates embeddings: Creates numerical representations (embeddings) of the text chunks using OpenAI's embedding model.
6. Inserts into Pinecone Vector Store: Stores the generated embeddings and their corresponding text chunks into a Pinecone vector database.
7. Loads Pinecone Vector Store: Retrieves relevant document chunks from Pinecone based on the user's query.
8. Retrieves relevant information: Uses a Vector Store Retriever to find the most relevant document sections for the given question.
9. Answers questions using OpenAI: Leverages an OpenAI Chat Model to generate an answer to the user's question based on the retrieved document content.

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running n8n instance.
• Google Drive Account: To store your PDF documents.
• OpenAI API Key: For generating text embeddings and AI-powered answers.
• Pinecone Account: To create and manage your vector store for document embeddings.

Setup/Usage

1. Import the workflow: Download the provided JSON and import it into your n8n instance.
2. Configure Credentials:
   • Google Drive: Set up a Google Drive credential to allow n8n to access your files.
   • OpenAI: Configure an OpenAI credential with your API key.
   • Pinecone: Set up a Pinecone credential with your API key and environment details.
3. Adjust Node Parameters:
   • Google Drive (Node 58): Specify the ID of the PDF file you want to process.
   • Pinecone: Insert (Node 1211): Configure your Pinecone index name and namespace if applicable.
   • Pinecone: Load (Node 1213): Configure your Pinecone index name and namespace, matching the insert node.
   • Manual Chat Trigger (Node 1207): This node acts as the entry point for your chat interactions. You will need to interact with it manually or integrate it with a chat platform (e.g., Slack, Telegram) using another trigger node.
4. Activate the workflow: Once configured, activate the workflow.
5. Interact with the workflow:
   • To upload and embed a new PDF, you would typically trigger the workflow with a specific command or input that leads to the Google Drive and Pinecone Insert path.
   • To ask a question, trigger the Manual Chat Trigger node with your question. The workflow will then retrieve relevant information from Pinecone and provide an AI-generated answer.