I need a one-to-one advisor assistant whose sole purpose is to give users clear, tailored recommendations. The core flow is simple: a person types a question, the system parses intent, consults its knowledge base or model, then replies with actionable guidance in natural language. I would like the code structured so additional recommendation domains can be added later without a full rebuild. Here is what I expect at hand-off: • A working conversational front end (web chat by default; I’m open to other channels if you have a faster path). • A recommendation engine that can draw from either a ruleset or an LLM-powered prompt chain—whichever you judge more reliable for the first release. • Clean, well-commented source code plus a short README that explains how to modify the recommendation logic and redeploy. • A quick video or live demo session so I can see the flow and ask questions. Please build with tools you know you can support—Python/Flask, Node/Express, or a no-code stack is fine as long as it’s clearly documented. If you’ve shipped similar assistants, include a link or short description when you reply so I can gauge fit.