## The everyday problem

Every department needs the voice of the customer, and every department gets a different version of it. Research writes a thirty-page report nobody outside research finishes. Product sifts beta surveys Monday morning under deadline pressure and goes with gut. Marketing pulls testimonial copy from whichever NPS responses sales happens to forward. Sales rebuilds the objection-handling deck from memory quarterly. Exec runs its own summary the night before the board meeting. Five teams, five summaries, five stories about the same customer base — and the gold under all of it (the actual transcripts) sits in a Drive folder nobody has time to read.

## What you'd type into the New Agent form

The head of UX research clicks **New Agent** in the dashboard, starts from scratch (no template matches research synthesis directly), and fills in:

**Agent name** · `Customer research synthesizer`

**Description** · `Turns transcripts and surveys into themed summaries with quote evidence. Same synthesis rules every team uses.`

**Persona** · `You are a senior UX researcher at [Company]. You synthesize interview transcripts and survey responses into ranked themes, each backed by direct customer quotes with timestamp citations. You weight responses by frequency, not by length.`

**Context to paste**

- The research framework (jobs-to-be-done, opportunity scoring)
- The standard interview question bank
- Two or three past research reports that hit the right structure

**Skills to pick**

- `transcript-synthesis`
- `theme-extraction`
- `quote-citation`

**MCPs to wire**

- Drive / Notion — where transcripts live
- Linear — where insights become opportunities
- The customer-research warehouse (NPS, CES, survey aggregates)

A few minutes of paste-and-pick. No code.

## What your team sees when they use it

Same agent, five consumer flows:

- **Research** runs interviews and gets v0 of the themed summary in fifteen minutes instead of two days. Every claim traceable to the source line.
- **Product** pastes 220 beta-survey responses; gets a ranked friction list in five minutes. Sprint planning moves from "what do we think" to "here are the top five things customers said."
- **Marketing** runs the agent on recent NPS free-text; gets customer-language phrases for ad copy and testimonials. Permission flags surfaced per quote.
- **Sales** runs lost-deal post-mortems through it; gets objection patterns that recur. The handling deck stops being a quarterly memory exercise.
- **Executive** runs quarterly NPS through it; gets the top three customer themes for the board. Same source as the roadmap.

## How it composes

This agent inherits the org defaults (company background, research workflow). On top, it overlays research-specific skills: transcript synthesis, theme extraction, quote citation. When the head of research tightens the canonical rules — weighting changes, a new framework — every team's next invocation picks up the corrected synthesis on next sync.

## How it evolves

Six weeks in, the team notices the agent over-weighting vocal customers (the 800-word responders get more representation than the 220 customers who left a single sentence). The head of research tightens the prompt to weight by frequency, not volume, and ships v2. **Every consuming team's next invocation picks up the corrected synthesis** — product's roadmap reads, marketing's testimonial pulls, sales' objection patterns, the CEO's board summary. Nobody has to be told to "use the new method"; it's the only method. Past summaries stay in the audit log — when a PM later defends a roadmap decision, they can show which prompt version produced the insight.