The agent assembled a reasonably complete picture of Cheesecake Labs by fetching the homepage, about, and services pages, but had to work around a JavaScript-heavy site that doesn't expose static content. Critical information like pricing details and case studies were either behind 404s or client-side rendering, forcing the agent to rely on prior knowledge and third-party sources (Clutch, G2, ZoomInfo) that were discovered via web search but not directly integrated into the final answer.

• ›Steps [1], [3], and [4] returned only HTML shell/meta tags with minimal readable content—the actual page content is loaded client-side by JavaScript, making static fetches uninformative. The agent could infer presence of pages (about, services) but not extract details from them.
• ›Pricing page ([5]), case studies ([7]), and work/portfolio ([8]) returned 404s, leaving gaps the agent could not fill from the site itself. The agent did not cite or extract pricing from these failures.
• ›The agent's final answer cites Clutch, G2, and ZoomInfo in the sources section, but those were discovered via web search ([10], [11], [12]), not fetched. The actual content used to answer the pricing and differentiation questions came from prior knowledge or search result snippets, not from cheesecakelabs.com itself.
• ›The site structure is not agent-friendly: no static content, no dedicated pricing/case studies URLs, no sitemap or structured data visible in fetched responses. The agent had to guess at URL patterns and resort to external sources.