You did 12 customer interviews this month. You have 12 Otter transcripts. And you probably listened to 6 of them and skimmed notes on the rest. The interviews were fine. Synthesis is where it falls apart.
Individual interviews are full of noise. The signal only shows up when you connect dots across 8 or more of them, and reading 12 transcripts by hand to build a theme map takes 6 hours you do not have. So it does not happen. The interviews go to waste.
The agent works in three passes. It reads every transcript, Otter, Fireflies, or manual notes, and pulls who was interviewed, the key quotes, the pains, the features they talked about. Then it extracts themes across all of them at once. Not "they mentioned onboarding" but "8 of 12 said onboarding was slow AND tried workarounds that failed AND took 2+ weeks to get production-ready." Then it turns those themes into hypotheses.
To run it you need Otter or Fireflies for the transcripts, Notion or Google Docs as the research repository, your CRM to map each interview to segment and company size, and your historical interviews so the agent can tell a new theme from a recurring one. Set it to run every Wednesday at 10 AM against the past week's calls.
The report runs six sections
One report, six sections: recurring themes (anything in 3+ interviews, with 2 to 3 quotes each), top pains vs. gains ranked by frequency and severity, feature mentions, testable hypotheses, contradictions between interviews, and segment insights across enterprise, mid-market, and SMB.
The one that earns its keep is testable hypotheses. The agent pushes each theme until it is specific enough to test. "We should make onboarding faster" is not a hypothesis. "Small engineering teams abandon API onboarding because our docs assume a dedicated integration engineer" is. And it keeps the quotes all the way through, so you hear the customer instead of a flattened summary of the customer.
One person saying onboarding is slow is an anecdote. Do not build a roadmap on it. Eight people saying it, across different company sizes and roles, with the same workaround pattern, is a signal. That threshold is the whole reason to automate this. No human reliably holds eight transcripts in their head at once and spots the pattern cutting across all of them. The agent does.
It catches the contradictions early too. "Enterprise loves our API, SMB finds it confusing" is a messaging fix, and you want it now, not next quarter.
Feed in your last 10 transcripts with segment metadata for each interviewee, run it, ask for the top three testable hypotheses. That output is your research backlog.
FAQ
What does the agent do with the output? It gives you synthesis in one place, actionable hypotheses specific enough to test, and conversation velocity so a new interviewer can read the synthesis instead of listening to all 12 recordings.
What data sources does it connect to? Otter or Fireflies, Notion or Google Docs, your CRM for segment mapping, and your historical interviews so it distinguishes new themes from recurring ones.