The translation of natural (human) language into SQL statements is nothing particular new. It exists now for quite a while and with the appearance of Large Language Models this functionality is becoming more and more visible.
Well, in most cases you have to live with a major draw back - the SQL will be generated somewehre in the cloud, not on your premise. Secondly and unfortunately, the Exasol database is not one of thre most widely used databases. The LLM for translation tasks do not see many of the Exasol SQL statements for training. Check out the article below on how
you can translate human language into functional SQL statements for the Exasol database, on-premise and therefore ‘safe’.
Just keep in mind, the quality depends on how you formulate yyour natural language question, the prompt engineering and the selected Large Language Model. While I am not generating SQL statements 100% correctly, the “hit” rate is really high and closing in to an acceptable value - still some work to do. I would say I am on the final approach phase…
Als always - Happy Exasoling.