Sviluppo di sistemi conversazionali per l’interrogazione di basi dati aziendali tramite linguaggio naturaleModern enterprises store vast quantities of structured information in relational databases, yet the ability to query that data remains concentrated among the few employees proficient in SQL. The result is a structural bottleneck: business users depend on technical intermediaries to obtain answers that are, in principle, directly expressible as queries. The problem of Natural Language to SQL (NL2SQL) translation, namely the automatic conversion of a natural language question into an executable SQL query, has long sought to remove this barrier, and the advent of Large Language Models (LLMs) has finally brought conversational NL2SQL systems within reach of production enterprise software rather than research prototypes alone.