n8n Workflow: Document Breakdown into Study Notes using Mistral AI and Qdrant

This n8n workflow automates the process of transforming raw documents into structured study notes. It leverages Mistral AI for intelligent summarization and question-answering, combined with Qdrant as a vector store for efficient information retrieval. The workflow is designed to process documents from a local directory, break them down into manageable chunks, generate study notes, and store them for easy access.

What it does

This workflow performs the following key steps:

1. Monitors Local Directory: It continuously watches a specified local directory for new or updated files.
2. Loads Documents: When a new file is detected, it loads the document content.
3. Splits Documents into Chunks: The document content is then broken down into smaller, more manageable text chunks using a Recursive Character Text Splitter, making it suitable for AI processing.
4. Generates Embeddings: Each text chunk is converted into a numerical vector (embedding) using the Mistral AI Embeddings model.
5. Stores Embeddings in Qdrant: These embeddings are then stored in a Qdrant Vector Store, allowing for semantic search and retrieval.
6. Generates Study Notes: For each chunk, it uses a Mistral AI Chat Model to generate concise study notes.
7. Summarizes Documents: It can also generate a summary of the entire document using a LangChain Summarization Chain.
8. Answers Questions: The workflow includes a Question and Answer Chain that can retrieve relevant information from the Qdrant vector store and answer questions based on the document content.
9. Processes in Batches: Documents are processed in batches to manage resource usage and API limits.
10. Handles Output: The generated study notes and summaries are processed and can be outputted for further use (e.g., saved to disk, sent to another service).

Prerequisites/Requirements

To use this workflow, you will need:

• n8n Instance: A running n8n instance.
• Mistral AI Account: An API key for Mistral AI to access their embedding and chat models.
• Qdrant Instance: Access to a Qdrant vector database (either self-hosted or cloud-based). You will need the API key and host URL.
• Local File Access: The n8n instance needs read access to the local directory where your documents are stored.

Setup/Usage

1. Import the Workflow: Download the provided JSON and import it into your n8n instance.
2. Configure Credentials:
   • Set up a Mistral AI credential with your API key.
   • Set up a Qdrant credential with your host URL and API key.
3. Configure the Local File Trigger: Specify the local directory path that the workflow should monitor for new documents.
4. Adjust AI Models (Optional): Review the "Embeddings Mistral Cloud" and "Mistral Cloud Chat Model" nodes to ensure the correct models and parameters are selected for your needs.
5. Customize Prompts (Optional): Modify the prompts within the LLM Chain nodes to tailor the generated study notes and summaries to your specific requirements.
6. Run the Workflow: Activate the workflow. It will automatically start processing files in the configured directory.

This workflow provides a robust framework for transforming unstructured documents into valuable study resources, making learning and information retention more efficient.