The agent successfully assembled a comprehensive understanding of Panta Insurance by relying almost entirely on web search results and external sources (Y Combinator, third-party review sites) rather than the company's own website. The homepage and most key pages on pantainsure.com were either inaccessible (404 errors) or returned truncated HTML that didn't contain meaningful content, forcing the agent to work around a largely non-navigable website and fill gaps with prior knowledge and external references.

• ›Direct fetches of pantainsure.com pages [5–11] all returned 404 errors or truncated responses, making the site's own content largely unusable. The agent discovered that /how-it-works, /pricing, /about, /general-liability, /commercial-auto, and /workers-compensation pages do not exist or are not indexable.
• ›The agent retrieved actionable content from web search results [13, 14, 21, 22] linking to Panta's Y Combinator profile and third-party review sites, plus the FAQ and coverage pages that did load. Searches about the company's business model, pricing, and differentiation were more productive than navigating the website itself.
• ›Y Combinator profile [19] provided critical context: founder names, founding year, licensing details, and the AI-native positioning—information not apparently available on the company's homepage or standard pages.
• ›The site published no transparent pricing information, service level details, or minimum account thresholds. The agent had to infer business model (commission-based) and target customers (high-risk industries) from scattered sources and Y Combinator context rather than clear messaging on the site.
• ›The agent worked around a nearly non-functional information architecture by pivoting to external sources and prior knowledge (64% of fetches came from prior knowledge, only 27% from the homepage). This indicates the website does not serve as the primary discovery or explanation surface for agents seeking to understand the company.