Build a Financial Documents Assistant using Qdrant and Mistral.ai

Financial Documents AI Assistant This workflow creates a sophisticated AI-powered assistant capable of answering questions about a collection of financial documents. It automates the process of keeping an AI knowledge base synchronized with a local folder of files, ensuring the assistant's responses are always based on the most current information. Technologies Used - n8n: The core automation platform coordinating the entire process. - Local File System: Used as the source-of-truth for the documents. - Qdrant: A high-performance vector database that stores the knowledge extracted from the documents for fast, semantic searching. - Mistral.ai: A Large Language Model (LLM) provider used for both creating vector embeddings and generating conversational answers. - LangChain Nodes: n8n's integrated LangChain nodes are used to structure the AI interactions, including text splitting, data loading, and creating the question-and-answer chain. How AI Integrates into the Flow This workflow uses AI in two critical stages of its Retrieval-Augmented Generation (RAG) architecture: 1. Document Indexing (Knowledge Creation): When a financial document is added or updated, its text is processed by the Mistral.ai Embeddings model. This AI model converts the text into numerical vectors (embeddings), capturing its semantic meaning. These vectors are then stored in the Qdrant database. This allows the system to find relevant information based on meaning, not just keywords. 1. Question Answering (Knowledge Retrieval & Synthesis): When a user asks a question via the chat interface, the Mistral.ai Chat Model acts as the brain of the operation. First, the user's question is also converted into a vector. The system then queries the Qdrant database to find the document chunks with the most similar vectors (i.e., the most relevant information). This context, along with the original question, is passed to the Mistral Chat Model, which synthesizes a comprehensive, human-like answer based on the provided financial data.