Turns a round of transcripts into a record where every quote can be checked.
Converts a round of interview transcripts into one record that opens with what the round was for and says whether it delivered it: every quote exact and carrying a locator, the prompting question recorded verbatim beside it, and how many were weighed against how many were kept. It will not work from notes or an AI summary, which are recollection with better handwriting.
Exists because the question that produced an answer is the difference between evidence and a planted finding. Asked about integrations and asked whether the integrations were painful compress to the same summary and are opposite evidence.
What it needs, and which agents use what it produces.