The agent successfully assembled a comprehensive explanation of Anvil Secure's business by fetching service pages, blog posts, and LinkedIn, but encountered significant friction: the site does not publish pricing information anywhere, forcing reliance on industry inference; key pages like /pricing, /services, and /about returned 404s, requiring web search to discover the correct URL paths (e.g., /about-us instead of /about); and the site's navigation structure is poorly indexed, necessitating 67% of fetches to come from web search rather than following onsite links. The agent delivered a complete answer to 3 of 4 requested dimensions (what, who, how different) but could only infer pricing from external context.

• ›Service pages (steps [8], [9], [6]) and blog posts (steps [12], [18]) were content-bearing and explicitly cited, providing detail on penetration testing, application/cloud security, embedded systems, AI security testing, and the '3 Basic Differentiators' positioning. These pages exist but are not discoverable via intuitive URL paths.
• ›Pricing is completely absent from the site — no /pricing page, no service menu with rates, no FAQ. The agent correctly identified this gap and noted that the contact form requires custom scoping. This is intentional opaqueness, not a technical issue, but represents a significant barrier to agent evaluation.
• ›The site's IA is broken: standard paths like /pricing, /services, /about all return 404s. The correct path is /about-us, but the homepage and navigation do not surface this. The agent discovered correct URLs via web search (step [4], [7], [19], [21]) rather than by following site navigation, indicating poor internal linking.
• ›LinkedIn (step [14]) and blog posts (steps [12], [18]) provided cultural/differentiator context that the main service pages did not fully articulate. The agent had to assemble narrative from fragments across multiple domains.
• ›Company identity was conflated: the domain is anvilsecure.com but internal references and blog posts use 'Anvil Ventures' — the agent noticed this confusion but had to infer the relationship from blog post metadata rather than from clear company documentation.