The agent successfully assembled a comprehensive overview of Fume's business model, target audience, and differentiation by combining content from multiple pages and web searches, but encountered significant friction around pricing disclosure and product boundaries. The site is partially agent-friendly — key pages load successfully but are JavaScript-heavy React apps that limit static content extraction, and critical information like pricing is either hidden behind non-functional URLs or requires prior knowledge to reconstruct.

• ›Content-bearing fetches (homepage [1], about [4], docs [10], blog [18], ai-agent-funds [27], omniclear [29], tokenization-engine [24]) returned valid 200 responses but the raw HTML was truncated React boilerplate with minimal visible text — the agent relied heavily on prior knowledge (33% of sources) and web search (40%) rather than parsing page content directly.
• ›Pricing page was completely inaccessible: search results advertised /pricing as a working page [12, 26], but fetches returned 404 [3, 15]. The agent documented this gap explicitly but had to infer AUM-based pricing structure and 0.5% max commission from prior knowledge rather than from retrieved content, indicating either obfuscated or broken pricing infrastructure.
• ›Product positioning was scattered across multiple pages (agents, tokenization engine, OmniClear, fund registry) with unclear boundaries; the agent had to synthesize these into a coherent narrative. No single page clearly articulated the full product stack or how pieces integrate — a major usability gap for evaluation-stage prospects.
• ›Web search proved more valuable than direct site navigation: 40% of sources came from search results, which surfaced blog posts and specific product pages that direct URL guessing missed (e.g., /resources/about, /ai-agent-funds, /omniclear). The site's IA is opaque to blind navigation.
• ›Integration and implementation details (setup time, admin software compatibility matrix, deployment costs) were not found anywhere, forcing the agent to mark them as missing — a critical gap for a B2B SaaS product targeting fund administrators.