Topic & Product Clusters

Use the otterly skill to extract and analyze how AI engines describe {BRAND} (and competitors: {COMP1, COMP2, …}) for brand report {report name or ID}, country {us}, dates {YYYY-MM-DD → YYYY-MM-DD}, engines {chatgpt, perplexity, gemini, google, google_ai_mode, copilot}.

Build the corpus. List the report's prompts (/prompts), then for each prompt pull /prompts/{promptId}/ai-responses. Keep: promptId, prompt text, engine, run date, response content, brandMentions, citations. Skip empty/failed runs.
Topic clusters — over the PROMPTS. Cluster semantically by intent, product category, use case, and funnel stage (TOFU/MOFU/BOFU, branded vs. unbranded). For each cluster output: name · prompt count · 3 exemplar prompts · avg. brand coverage · engines where {BRAND} wins vs. loses.
Product characteristics — over the ANSWERS. For every response, extract sentences where {BRAND} (and each competitor) is named. From those passages, mine: descriptive attributes (adjectives, qualities), positioning claims, use cases, target audience, price/quality signals, head-to-head comparisons, recurring caveats/weaknesses. Aggregate into a profile per brand: attribute → frequency → engine breakdown → 2–3 verbatim quotes (≤25 words, with engine + date).
Synthesis. Deliver:
a. one-paragraph executive summary,
b. topic cluster table,
c. brand characteristic profile,
d. side-by-side perception matrix vs. competitors,
e. 3 actionable observations (strongest perception, weakest, biggest gap vs. desired narrative).

Tip: replace anything in [BRACKETS] with your own values, then paste this into Claude with the OtterlyAI connector enabled.