The agent assembled a comprehensive explanation of FF Forest by combining web search results with third-party review sites, since the official website itself returned only JS-rendered shells. The site's homepage and key pages (/en/about-us, /en/how-it-works) were technically reachable but not content-bearing in the fetches — the agent had to rely on external sources (northern.finance review, marcoschwartz.com review, CrowdSpace platform overview) to extract the actual business model, pricing, returns, and risk information needed to answer the task.

• ›Steps [7], [11], [12], [14], [15] were web searches that surfaced external review and platform-analysis sites; steps [8], [9], [10], [13] fetched those external sources. The official FF Forest pages ([8], [15] fetch) returned only Next.js app shells with no meaningful content in the HTML, forcing the agent to rely entirely on third-party sources for substantive information.
• ›The agent cited 8 sources in the final response; 6 of them ([8], [9], [10], [13], [15], and search results leading to them) came from external domains (northern.finance, marcoschwartz.com, thecrowdspace.com, p2pdash.com, finforta.com). Only 2 citations ([8] 'FF Forest About Us' and [15] 'How It Works') pointed to the official site, but those pages were not content-bearing in the actual fetches.
• ›FF Forest's site is not agent-ready for discovery: standard routes like /pricing and /about returned 404s; the homepage and main pages are client-rendered SPAs that serve no text content in the initial HTTP response. The agent had to infer the business from external reviews and competitor/platform analysis sites, meaning the official site provided almost no machine-readable information about what FF Forest does, its pricing, or how it works.