The agent assembled a reasonably complete picture of Falcon Copper by combining web search results and external sources, since the site itself is heavily client-rendered and did not expose full content in HTML fetches. The agent successfully answered what Falcon does (integrated US copper supply chain), who it's for (government, investors, large industrial buyers), and how it differs from competitors (vertical integration, new smelting capacity, strategic partnerships). Pricing and revenue model remain partially unclear because Falcon is not a service with listed rates—the agent inferred a tolling model but noted this was not explicitly stated on the site.

• ›The site's Next.js architecture delivered only meta tags and script bundles in HTML; full page content was client-rendered and inaccessible to fetches [1, 3, 7, 18, 19]. The agent had to rely on web search snippets and external sources to answer the task.
• ›Search results [11, 15, 16, 21, 22] contained sufficient snippet text to infer business model, partnerships, and differentiators—particularly the Glencore and Kennecott references and the mention of primary smelting projects. These were the actual information sources, not the site itself.
• ›The agent correctly identified that Falcon is a capital-intensive mining/industrial company, not a SaaS vendor, and therefore correctly noted that 'pricing' (as a traditional service offering) does not apply. However, this insight came from prior knowledge and search context, not from site content.
• ›Timeline, project status, monetization mechanics, and target customer specifics were explicitly flagged as missing or unclear—information the site did not surface to either HTML fetches or search indexing.
• ›The site is poorly navigable for agent discovery: fetches returned truncated responses, expected pages like /pricing returned 404s [4, 5], and no internal links or structured data were exposed to guide the agent to content-bearing resources.