Business analysts and data consumers who need ad-hoc answers from complex data warehouses must translate ambiguous natural language questions into correct, optimised SQL — a task complicated by schema complexity, dialect differences, and hidden business logic embedded in table relationships.