The agent successfully gathered comprehensive information about Conductor by combining structured markdown exports (llms.txt, pricing markdown, team page) with external web search to contextualize alternatives. The site's machine-readable markdown layer (/llms.txt, /markdown/* routes) made core facts discoverable, but comparison positioning and some feature details required external sources, indicating the site prioritizes product narrative over competitive clarity.

• ›Steps [6, 8, 9, 15] returned cleanly parseable markdown with product definition ('Mac app that lets you run many coding agents in parallel'), pricing tiers ($0/$50/$60+), and team context — all directly usable. The /llms.txt probe was the linchpin, directing the agent to /markdown/* variants of key pages.
• ›Step [14] (web search) was the only source for competitive differentiation (vs. Pane, Cursor, etc.). The site itself contains no comparison pages or 'why Conductor' positioning, forcing the agent to assemble context externally. This gap is acknowledged in the final response's 'What Was Confusing' section.
• ›The agent had to work around limited inline documentation: pricing pages lack detail on future usage-based billing, Team plans redirect to a contact form, and 'Conductor Cloud Computer' is mentioned but unexplained. HTML responses ([1–5, 7, 12–13]) contained rendered SPA content the agent could not extract; the agent pivoted to /markdown/* and /llms.txt routes, which were machine-readable and effective.
• ›The site's dual-layer strategy (rendered marketing + markdown exports for LLMs) works but is fragmented. Pricing details, team benefits, and differentiators required synthesis across multiple markdown files plus external search. A single 'llms.txt' or comparison-focused page would have been more direct.