Generative artificial intelligence based conversion of natural language requests to data warehouse query instruction sets

A system for converting a natural language query to computer readable instructions for generating computer readable instructions to query a data warehouse, the system comprising:

• one or more processors;
• an interface;
• one or more database query tools;
• and
• a memory storing:
• a table description of a plurality of tables stored in a remote data warehouse;
• and
• instructions that, when executed by the one or more processors, causes the system to perform operations comprising:
• obtaining, by the interface, a natural language query as an input for querying the remote data warehouse;
• iteratively generating, using one or more database query tools in conjunction with a generative artificial intelligence (AI) model, a final query instruction set for querying the remote data warehouse from the natural language query based on the stored table description, wherein the one or more database query tools are prevented from accessing the remote data warehouse to generate the final query instruction set, the iteratively generating including, over a first iteration and at least one next iteration:
• generating a context from a current input using one or more database query tools based on the stored table description, wherein the natural language query is the current input for the first iteration of the iteratively generating;
• providing the current input and the context to the generative AI model;
• receiving an intermediate input in a natural language from the generative AI model based on the current input and the context;
• determining whether a threshold number of iterations or a threshold amount of time is reached, wherein determining whether the threshold number of iterations or the threshold amount of time is reached is performed concurrently with generating the context from the current input;
• and
• selectively ending the iterative generation based on whether the threshold number of iterations or the threshold amount of time is reached, the selective ending of the iterative generation including:
• ending the iterative generation responsive to determining that the threshold number of iterations or the threshold amount of time is reached, wherein the final query instruction set is generated based on the intermediate input;
• and
• starting the next iteration responsive to determining that the threshold number of iterations or the threshold amount of time is not reached, wherein the intermediate input is used as the current input for the next iteration;
• and
• outputting, via the interface, the final query instruction set.