I Turned 27 Business Documents Into a Private AI Knowledge Base

What if your contracts, invoices, price lists, procedures and technical manuals could become one private AI knowledge base running locally on your own machine? In this project, I turn 27 real-world business documents into persistent, searchable knowledge that can answer operational questions, compare supplier information, calculate pricing, connect technical relationships and always point back to the original evidence. Business automation & software integrations: https://levixflow.com This goes beyond a basic “chat with your PDFs” RAG demo. The system combines local AI, structured business facts, semantic retrieval, relationships and source provenance so different questions can use different retrieval strategies instead of sending everything to an LLM. In the video, I show how the system can: • answer questions across contracts, invoices and price lists • calculate quantity-based pricing using the correct commercial terms • compare supplier information across multiple documents • answer technical questions from product documentation • show the original source and evidence behind important answers • build persistent, reusable business knowledge • visualize connected knowledge using Obsidian • run completely locally and offline • ingest a new document and make the new information immediately reusable • refuse to guess when the available evidence is insufficient The local stack includes Ollama, Qwen 3.5 4B, Qdrant, SQLite, FastAPI, Streamlit and a Markdown knowledge layer. The architecture uses ideas from Retrieval-Augmented Generation (RAG), semantic search, private AI knowledge bases, structured retrieval and query routing, but the goal is not to build another AI chatbot. The goal is to turn scattered business information into persistent knowledge that people can actually use. AI understands the documents. The system remembers. If you work with ERP systems, ecommerce, spreadsheets, documents, back-office workflows or business automation, you can find more of my work at Levixflow. #PrivateAI #RAG #KnowledgeBase #LocalAI #BusinessAutomation #Ollama #Qwen #SemanticSearch #EnterpriseAI #DocumentAI