Hey guys, just had a situation where the instruction to AI enrichment was run only on cells that haven't run yet or have errors and it re ran a bunch of cells that did not meet that condition causing tons of duplicate api credit usage. Any idea why?
This could've been because of some potential changes in the prompt, the API considered that as a new request so it re-ran all of them :/... I'll DM you.As a quick tip, if you think the final result will not change much even after modifying the prompt/column a bit. Then I would apply a filter to show those that don't have any results yet and then run them