Internal AI assistant for knowledge base (RAG) with citations

Our team loses hours every week searching for answers that already exist somewhere but are scattered across documents, wikis, and PDF files. I want to hire a specialist to build an internal AI assistant for a knowledge base using retrieval-augmented generation. The system should load company documents, split them into chunks, build embeddings, and store everything in a vector database. When an employee asks a question in the chat interface, the assistant finds the most relevant chunks and formulates a well-reasoned answer with clear links to the source document and section. This is specifically an internal Q&A tool for documents, not a customer support chatbot. Accuracy and traceability are more important than chattiness. Every answer should show where the information came from so an employee can verify it, and if the data is missing, the assistant must honestly say it doesn't know rather than make something up. I care about careful handling of document updates so the index stays fresh, reasonable response latency, and access control so only authorized employees access sensitive materials. Describe which embedding model, vector store, and orchestration approach you'll suggest and why. In the result, I expect a working chat interface, a repeatable document loading pipeline, and brief documentation so our team can maintain the system themselves. In your response, show similar RAG projects and estimated timelines. — Loading pipeline: parsing, chunking, and embedding documents into vector storage — Chat interface where employees ask questions and get answers with source links — Quality tuning and a way to update the index as documents change — Access control, logging, and brief maintenance guide for our team