I need a text-based chatbot that can ingest PDF, Word, and Excel files, build an internal knowledge base, and then respond to user questions or requests for summaries in real time. The core tasks are straightforward: • Search & retrieval – a user should be able to ask natural-language questions and immediately receive answers that cite the relevant passages, cells, or pages. • Content summarization – on request, the bot should generate concise summaries of an entire file or a specific section the user highlights. Interaction happens only through text, so no voice commands or button workflows are required. A simple web interface, desktop app, or command-line tool is acceptable as long as it supports drag-and-drop or file-picker uploads and returns responses quickly. Deliverables • Source code with clear setup instructions (Python with LangChain, Node.js, or another modern stack is fine as long as dependencies are documented). • A lightweight UI that lets me upload or select multiple documents at once, enter questions, and see answers or summaries. • README that explains how to add more documents and retrain or re-index the model. • Short demo video or screenshots showing the chatbot answering at least three search queries and producing one summary drawn from a mix of PDF, Word, and Excel files. Acceptance criteria 1. Answers reference the correct location in the original document. 2. Summaries stay within 10% variance of the original word count limit I set (e.g., “200-word summary”). 3. Response time for a single query on a 10-document corpus (under 25 MB total) is under 8 seconds on a mid-range cloud VM. If you already have experience with vector databases (Pinecone, FAISS, or similar) or have built RAG pipelines with OpenAI or Azure OpenAI, mention it in your proposal along with a brief outline of your approach.