That’s good. Also maybe an architecture that runs the query through multiple times and then evaluates similarity of responses, then selects (or creates) the mos

That’s good. Also maybe an architecture that runs the query through multiple times and then evaluates similarity of responses, then selects (or creates) the most-generated one, along with a confidence level of how many of the individual responses were aligned.