🎧

Synthesise user interviews into insights

Turn raw, messy interview notes into evidenced pains and themes — with quotes, not vibes.

The job here is pattern-finding across messy text, with traceability back to evidence. Big-context models shine when you have many transcripts at once; a balanced model is fine for a handful.

Gemini

Gemini 3.7 Flash

Why this variant: When you have a dozen transcripts, hand it the whole pile rather than summarising each first — the huge context window means you skip the pre-trimming step entirely, which is faster and cheaper.

Prompt for Gemini 3.7 Flash

{paste all interview transcripts, separated by ---}

Across ALL transcripts above: list the top 5 recurring user pains. For
each, give one verbatim quote as evidence, which interview it came from,
and how many interviews mentioned it. Ignore anything that appears only
once. Table only.

Claude

Claude Sonnet 5

Why this variant: Ideal for a handful of interviews where you want tight, well-evidenced output. Tags keep the notes and the task separate; the format constraint keeps it from rambling.

Prompt for Claude Sonnet 5

<notes>
{paste raw interview notes}
</notes>
Pull out the top 5 user pains. For each: a one-line quote that evidences
it, and roughly how often it came up. Table only, no preamble.