Intelligent Document RAG with Mistral OCR and Gemini AI

• End-to-End RAG System: Seamlessly integrates OCR, embedding, vector storage, and AI response generation.
• Advanced Document OCR: Utilizes Mistral AI's mistral-ocr-latest model for highly accurate text extraction from PDFs.
• Intelligent Knowledge Retrieval: Stores and retrieves document chunks with OpenAI embeddings in a Qdrant vector database.
• Conversational AI: Powers interactive Q&A sessions using Google Gemini 1.5 Flash through a Langchain agent.
• Automated Document Ingestion: Streamlines the process of converting raw PDFs into an queryable knowledge base.

To get started with this powerful workflow:

1. Set up Credentials: Ensure you have the following n8n credentials configured:
   • Mistral Cloud account for Mistral AI API access.
   • OpenAi account for OpenAI embeddings.
   • Google Gemini(PaLM) Api account for Google Gemini LLM.
   • A Qdrant API credential (e.g., Qdrant API (Hetzner) / QdrantApi account (Hetzner)) for your vector database.
2. Configure Qdrant URL: In the Refresh collection HTTP Request node, update http://QDRANTURL/collections/COLLECTION/points/delete with your actual Qdrant instance URL and collection name. Ensure the collection name ocr_mistral_test (used in Qdrant Vector Store1 nodes) is consistent.
3. Initiate Document Ingestion: Trigger the upper workflow path (starting with When clicking ‘Test workflow’) to upload your PDF documents to Mistral AI for OCR processing, embedding, and storage in Qdrant.
4. Engage with the AI Chatbot: Once documents are ingested, use the When chat message received trigger to send queries. The chatbot will retrieve relevant information from your documents and provide answers via Google Gemini.