The agent assembled a workable explanation of The Token Company's business by relying almost entirely on external third-party sources (Y Combinator, PointFive guides, and Respan alternatives page) rather than the company's own website. The site itself—rendered heavily in JavaScript with minimal extractable content and missing key pages (/product, /about returning 404s)—was effectively unnavigable for this discovery task; the agent had to work around it by searching outbound.

• ›The agent's core information about what The Token Company does, pricing ($0.05/1M tokens), who it's for, and competitive positioning came entirely from steps [15] (Y Combinator profile), [16] (PointFive comparison guide), and [20] (Respan alternatives map)—not from thetokencompany.com itself.
• ›Attempts to fetch the site's own pages (/, /pricing, /docs, /blog, /product, /about) returned either JavaScript-heavy HTML that was not content-bearing or 404s; the agent could not extract usable information directly from the domain despite multiple fetch attempts.
• ›The agent explicitly flagged in its final response that 'The pricing isn't prominently displayed on the pricing page itself—I had to find it in the page's meta description' and that '/product and /about pages don't exist (404s)', revealing that the site's own navigation and documentation are incomplete or inaccessible.
• ›The site's JavaScript-first architecture prevented the agent from parsing detailed content; the agent resorted to web search (steps [13] and [18]) and citations of external sources as a workaround, indicating poor agent-readiness.
• ›The agent's final answer is accurate and complete, but it was synthesized from alternative sources (Y Combinator, guides, market maps) rather than the company's own website, suggesting The Token Company does not self-publish its value proposition, pricing, or competitive differentiation in a machine-readable or easily navigable form.