The agent successfully gathered comprehensive information about WriteHuman by fetching the homepage, pricing page, and about page directly from the site, then supplemented with web search results to fill gaps and verify competitor positioning. The site's main pages were navigable and returned structured content, but the agent had to rely on third-party reviews and search snippets to fully answer questions about differentiation, target audience, and feature details that weren't explicitly stated on writehuman.ai itself.

• ›Steps [1], [2], and [7] (homepage, pricing, and about pages) were the primary content sources that returned actual site HTML; the agent extracted core positioning, pricing tiers, and feature claims from these fetches, though the HTML itself was not fully parsed in the trajectory view.
• ›The agent used web search results (steps [4]–[6], [9]–[11]) as the main source for competitive differentiation, use cases, audience targeting, and feature clarification (e.g., 'Enhanced-Mode Cliff', language support, MCP access). This indicates WriteHuman.ai's own site does not comprehensively document how it compares to alternatives or detail all audience segments.
• ›Pricing information was retrievable from the pricing page [2], but the agent noted conflicting pricing ($18 vs $22 for Pro, unclear Ultra tier details) and lack of clarity on team/enterprise billing — suggesting the pricing page itself may have been ambiguous or the agent had to reconcile multiple sources.
• ›The site lacks transparent documentation of company background, support channels, feature distinctions (Standard vs Enhanced mode trade-offs), and the meaning of 'MCP Access' — all gaps the agent flagged and had to work around using third-party reviews.
• ›The agent's final response cites 'WriteHuman Official Site', 'Pricing', and official pages, but the bulk of the structured comparison and use-case detail came from external review sites (WriteHybrid, Unite.AI, UndetectedGPT), indicating writehuman.ai itself does not publish this analysis natively.