Amazon AI Search Analyst

Problem

When researching demand outside Amazon, you often need Google AI Overviews, organic results, Reddit/Quora pain points, and keyword trends at the same time. Manual searches can miss exact citations, conflate relative trend scores with absolute search volume, or mix Amazon on-site product search with off-site SERP work.

How It Works

The skill works through MCP tool calls rather than local scripts or deep site crawling. Core capabilities include:
- Using ai_search in overview mode to fetch standard SERP items, AI Overview text, and citation sources.
- Using ai_search in ai_mode for multi-turn follow-up research when the user explicitly asks for deeper digging.
- Using keyword_trends to compare keyword popularity, geo heat, seasonality, and Breakout terms.

The key workflow is to classify the question first, then choose one overview call, a bounded ai_mode run, or a trend comparison. Output should be structured: titles, URLs, citations, trend notes, not raw JSON.

Boundaries

It only targets Google search result pages and keyword trends. It does not perform Amazon on-site product search, review scraping, niche/category filtering, or deep crawling of target websites. AI Overview may not appear for every query; fall back to organic results. keyword_trends returns 0-100 relative values, not absolute search volume. Multi-turn search is slower, so keep followups small and disclose latency before running it.