Abstract: In an example, a method is performed by sending query input and a first instruction to a generative machine learning model (GMLM). The first instruction is to cause the GMLM to generate and output a query input summary. The query input summary and a second instruction are sent to the GMLM. The second instruction is to cause the GMLM to provide query results using multiple different queries and the query input summary. The query results and a third instruction are to cause the GMLM to summarize a comparison of a query result with the query input summary. A query result evaluation summary is generated and output by the GMLM. A signal is received from a position within a presentation of the query result evaluation summary. The first instruction is updated to include the signal in the query input summary.

Abstract: An example sends query input and a first instruction to a generative machine learning model (GMLM). The first instruction is to cause the GMLM to generate and output a query input summary. The query input summary and a second instruction are sent to the GMLM. The second instruction is to cause the GMLM to provide query results using multiple different queries and the query input summary. The query results and a third instruction are to cause the GMLM to summarize a comparison of a query result with the query input summary. A query result evaluation summary is generated and output by the GMLM. A signal is received from a position within a presentation of the query result evaluation summary. The first instruction is updated to include the signal in the query input summary.

Abstract: An example provides a multi-agent system. Via an orchestrator agent, a query input is received. Via the orchestrator agent, the query input and a first instruction are provided to a generative machine learning model (GMLM). The first instruction is to cause the GMLM to determine queries using the query input. Via a query execution agent, the queries execute in parallel. Via the orchestrator agent, it is determined whether the queries are executing. Via a query evaluation agent, query results of execution of the queries and a second instruction are provided to the GMLM. The second instruction is to cause the GMLM to generate, for each query result, a query result summary. Via the orchestrator agent, the query result summaries are used to determine a subset of the query results for presentation via a device.