Claude Codeanalyze-interview

analyze-interview

Turn an interview’s responses into a grounded findings report.

Triggers on: “what did people say about X”, “summarize the interview results”, “how many finished it”.

What it does

  1. Identifies the deployment (via ear3_list_interviews if you don’t have the id).
  2. Pulls responses with ear3_list_responses — status, participant, AI summary; transcripts on request. Partial and completed sessions are counted separately, never silently mixed.
  3. Asks targeted questions with ear3_ask_insight — Ear3’s insight model answers over all responses with a summary, confidence, and supporting quotes. One call per distinct question.
  4. Reports, grounded:
    • Participation: total, completed vs partial.
    • Key themes with 1–2 short verbatim quotes each (never invented or paraphrased as verbatim).
    • Notable outliers.
    • If the data is thin, it says so plainly rather than over-reaching.

Then

Want the raw data instead of a summary? → export-responses.

Backed by the ear3_list_responses, ear3_ask_insight, and ear3_get_session tools.


Built by Ear3 — voice interviews for any app.
⌘/