The agent gathered partial information about Learnr by fetching the homepage and pricing page, but could not access the actual content of those pages due to React prerendering—all fetches returned only skeleton HTML with no visible text. The agent worked around this by assembling an answer from prior knowledge (60% of sources), web search results about AI course generators and gamification, and inference, ultimately delivering a plausible but largely unverified explanation of what Learnr does. The site is poorly navigable for agent-based discovery because it serves prerendered HTML shells to crawlers while hiding all substantive content behind client-side JavaScript rendering.

• ›Steps [1] and [2] returned only HTML boilerplate with no rendered content—the meta description and prerendering guards confirm the site relies entirely on React client-side rendering, leaving fetchers with only structural markup.
• ›The agent could not extract pricing details, feature descriptions, or positioning language from any of the 14 direct fetches to getlearnr.com URLs ([1], [2], [3], [4], [10], [11], [12], [14]); all returned identical prerendered shells with visibility:hidden applied to the root div.
• ›The agent bridged the gap using 60% prior knowledge (likely cached info about Learnr's positioning and creator Dev Sanghavi) and 8 web searches that returned third-party aggregator pages (Capterra, GetApp, etc.) and comparison guides—none of which contained substantive Learnr details either.
• ›Key facts in the final answer ('Duolingo for Everything,' XP/hearts/streaks, Dev Sanghavi, Houston, June 2026) appear to come from prior knowledge, not from the actual site fetches, since no fetched HTML contained this text.
• ›The agent explicitly flagged critical gaps: no visible Pro pricing, no free-tier limits, no content quality details, no Product Hunt presence—all of which reflect genuine absence from the fetchable content, not agent error.
• ›The site is fundamentally agent-hostile: it uses prerendering guards (`html.learnr-js #root[data-prerendered=true] { visibility: hidden }`), no fallback text content, and makes no effort to publish pricing or positioning in meta tags, JSON-LD, or static HTML.