What problem does it solve? Analyzing user research interviews manually is slow and error-prone: quotes get paraphrased, evidence goes uncited, and single-interview observations get presented as cohort findings. This Skill produces a rigorous, evidence-grounded per-interview write-up where every claim traces to a verified transcript line. ## Core Features & Use Cases - Verified findings pipeline: Writes every finding to BigQuery via write_finding with mandatory verbatim quote evidence, validated line IDs, and honest confidence scores; findings are always written as proposed for human review. - Pass-based transcript reading: Surveys tag summaries first, walks line ranges with compact notes, and verifies every citation with a dedicated tool before writing, with measured coverage reporting. - Structured field notes: Produces a sub-500-word Google Doc with 3-4 themes (each resting on 2+ tagged lines), sentiment analysis, verbatim quotes with figurative-language flags, and open questions or hypotheses typed distinctly. - Use Case: After a Tagger pass has tagged a two-hour user interview, run this Skill to produce a Field Notes document in the client's Drive folder, with each theme linked to BigQuery finding rows a reviewer can approve in Stu. ## Quick Start Analyze the tagged transcript for conversation 1042 in the client dataset and write the field notes document to the client Drive folder.
