The agent successfully compiled a comprehensive explanation of Amplitude's product, pricing, target audience, and competitive positioning by combining direct website fetches (which returned raw HTML with limited readable content) with web search results and third-party review sites. The official Amplitude site proved difficult to parse due to heavy JavaScript rendering, forcing the agent to rely primarily on external sources (UserPilot, PrettyInsights, Nasdaq, and BigDATAwire) to extract substantive information about features, pricing tiers, and differentiators like the Behavioral Graph and AI agents.

• ›Steps [1], [2], [4], [7], [8], [9], [22], [23] were fetched directly from amplitude.com but returned minimally readable HTML (truncated at 4400 chars) due to Next.js client-side rendering—the agent could not extract structured content from its own domain.
• ›Steps [11], [12], [19], [20] (web searches) returned snippet-based results that pointed to third-party review and news sites; these search results themselves contained enough summary information to seed the agent's understanding.
• ›Steps [15] (UserPilot) and [17] (PrettyInsights) were the only external fetches that returned full page content; the agent cited both in its final sources, indicating they were used for pricing details, alternatives comparison, and feature descriptions.
• ›The agent had to supplement site-sourced information with prior knowledge (~50% of fetches sourced from prior_knowledge per metadata), suggesting Amplitude's public-facing site does not comprehensively answer discovery-stage questions in machine-readable form.
• ›Pricing information was scattered: free tier details required parsing search snippets; Growth/Enterprise pricing transparency was explicitly called out as missing, indicating the site intentionally gates this information behind a sales contact requirement.
• ›The agent correctly identified and documented friction points (JavaScript rendering, pricing opacity, feature complexity) in its final response, demonstrating honest assessment rather than papering over gaps.