The agent successfully gathered enough information to explain ex.co's core business, target audience, and differentiation, though pricing details remained opaque by design. The site proved difficult to navigate directly (heavily minified, no public pricing page), so the agent relied on a press release blog post and an LLM-specific info document (llms.txt) to construct its answer. The agent assembled a coherent explanation from fragments rather than finding a single comprehensive source.

• ›Step [22] (ex.co/llms.txt) was the single most valuable resource—a structured, machine-readable document explicitly intended for LLMs that directly answered 'what do they do,' 'who it's for,' and core differentiators. This page was cited and heavily sourced the final answer.
• ›Step [21] (blog press release about LLM-based video recommendation engine) provided concrete product features and performance claims (49% engagement lift, 17% yield lift) that supported the differentiation narrative.
• ›Step [19] (web search) surfaced the llms.txt URL and press release—the agent did not discover these via site navigation but via search. The homepage and about pages returned heavily minified JavaScript with no readable content.
• ›The site deliberately hides pricing (no numbers on pricing-plans page, enterprise negotiation model), making cost comparison impossible—this is intentional but limits agent utility for price-sensitive evaluation.
• ›The agent had to assemble the answer from press releases, an LLM-specific info document, and web search snippets rather than from a cohesive public website. Site structure and navigation are not agent-ready; direct fetches of main pages (/, /about, /ovp, /pricing-plans) all returned unreadable minified code.