Build an AI Chatbot That Answers Questions From Website Content

We are looking for a developer to create a lightweight AI chatbot that answers user questions based on a predefined collection of website content or documents. The goal is to build a small proof of concept demonstrating retrieval-augmented generation (RAG), with an emphasis on useful answers grounded in the supplied content. Core Features A simple chat interface with conversation history during the active session. Support for a limited collection of supplied text documents or website content. A basic content ingestion and indexing workflow. Retrieval of relevant information before generating an answer. Responses generated through an LLM API. Clear handling of questions that cannot be answered from the available content. Basic error handling and a responsive interface. Preferred Technology Stack Next.js and TypeScript for the frontend, with a suitable backend and a lightweight vector search solution. A managed vector database or local alternative can be proposed if it fits the budget. Deliverables A working chatbot MVP. A demonstrable question-and-answer workflow using the supplied content. Source code in a Git repository. Setup instructions and API configuration documentation. Basic testing using a small set of representative questions. Budget: $100 fixed-price. The scope is intentionally limited to a small proof of concept. Large-scale crawling, complex multi-tenant architecture, authentication, analytics, production-grade infrastructure, and advanced agent workflows are excluded. Please describe your experience with RAG, embeddings, document retrieval, and LLM APIs. Include your proposed approach, estimated delivery time, and any third-party service costs separately.