Design a RAG-Based Support Assistant

A retrieval-augmented support assistant is the generative question interviewers now reach for: an LLM answering customer questions from a company knowledge base, where the hard parts are retrieval quality, hallucination control, evaluation without a single right answer, and per-token economics instead of per-request QPS math. This walkthrough follows the Interview Framework for a support assistant at a B2B SaaS company, with one announced deviation from the phase order that the framework itself prescribes for generative systems.