The agent assembled a partial understanding of hernigonz.com by combining prior knowledge about the founder with a single web search result (step [9]) that linked to Hernán González's LinkedIn profile. The site itself proved largely opaque: all direct fetches returned Next.js client-side-rendered HTML shells with no visible content, forcing the agent to rely on external search results and educated guesses about the platform's purpose. Critical information—pricing, specific course titles, delivery format, and detailed offerings—could not be retrieved from the site itself.

• ›Step [9] was the only content-bearing source: a LinkedIn profile for 'Hernán González - Entusiasta de Bitcoin' that established the founder's identity and background in Bitcoin education in Argentina. All other search steps returned generic or irrelevant results (Coursera, Udemy, other unrelated platforms).
• ›The site's direct fetches (steps [1], [3], [5], [8]) returned only HTML boilerplate with Next.js script tags and no rendered content. The platform uses aggressive client-side rendering, making it completely opaque to server-side inspection. The agent could infer page structure (routes like /cursos, /directorio, /contacto) but not what those pages contain.
• ›The agent filled knowledge gaps by inferring from the founder's LinkedIn profile (Bitcoin Argentina volunteer, LABITCONF involvement, Escuelas y Bitcoin) that the platform targets Spanish-speaking Latin American learners interested in Bitcoin/fintech. This was an educated assembly, not a retrieval from the site itself.
• ›Pricing, course names, course format, and certification details—core elements needed to fully answer the task—remain completely inaccessible. The agent explicitly called this out as a critical missing piece and recommended visiting the site directly.
• ›No structured data (schema.org, JSON-LD), meta tags, or public API endpoints were available to help the agent. The sitemap probe (step [19]) returned a 404.