Build a smart document-based AI assistant like Humata AI, but specific to my needs: ✅ Only I (the admin) should be able to upload or replace documents. ✅ The users should only be able to ask questions, nothing more. ✅ The documents I upload will be internal directives from the Ghana Civil Aviation Authority (GCAA) — covering all GCAA directives, not just RPAS. ✅ The AI must read and understand the uploaded documents, process them using embeddings and vector storage (e.g., Chroma or FAISS). ✅ Use LangChain, FastAPI backend, and OpenAI API (or a local LLM) to build a full Retrieval-Augmented Generation (RAG) system. ✅ When I upload new documents, the vector store should automatically refresh and the AI should now use the new documents to answer. ✅ The assistant should give detailed, intelligent responses based only on the uploaded documents. ✅ Do not cite me as a source — the assistant should "think" for itself based on the data. ✅ The front-end should be styled and ready — user input box only. No document upload for public users. Deliver the complete solution (backend + frontend) ready to run locally or deploy. - Initial Deployment
