An LLM-based router classifies analytical intent, dispatching either NL2SQL generation or Pandas-based exploratory workflows.
Controlled execution ensures generated analysis runs safely while still enabling rapid visualization and iteration in a Streamlit interface.
