User feedback clustering and themes
User feedback arrives through support tickets, app-store reviews, sales notes, and a community forum, and every quarter someone spends a week turning that pile into themes. Design an agent that clusters raw feedback into themes with representative quotes and rough counts, refreshed weekly. Tell me how it ingests each channel, how a human validates the themes before they steer the roadmap, and how you'd catch it inventing patterns that aren't there. Close by convincing me this deserves a spot on the team's budget.
select_supported_themes(theme_names: list[str], quote_counts: list[int], distinct_sources: list[int], min_quotes: int) → list[str][["slow sync","confusing pricing","crash on upload"],[12,5,3],[3,2,1],4]out["slow sync","confusing pricing"][["export bug"],[9],[1],3]out[][[],[],[],2]out[]Turn the fuzzy goal into a bounded agent task. Say what the agent reads, what it may do on its own versus draft for a human, and what it must never touch. Then earn the trust: how you’d verify it before rollout, and the pitch (the problem, what you built, why it matters).
Agent build: design the agent in plain language (or narrate it). The coach grades the decomposition, the guardrails, the verification plan, and the pitch.