applied-gen-ai/txt2sql ? reverse-engineered prompt

Build me a simple text to SQL demo that lets someone ask questions about a customer dataset in plain English and get the answer back from the data.

I want it to load a small sample CSV, understand the table schema, and route questions to the right helper so it can turn the question into SQL, run it, and show the result in a friendly way. It should include a little bit of smart matching for messy user wording, plus a basic knowledge base so it can recognize common customer terms and fields. If it helps, you can use notebooks or small Python scripts for the main flow, and make sure the examples are easy to run end to end.

Keep it focused on a lightweight prototype, not a full product. Add clear sample questions, sensible defaults, and a simple way to test the query flow. If you need current docs for anything, look them up online.