The agent successfully gathered enough information to explain AHT Production's business model, target customers, service areas, and portfolio, but worked around a fragmented website where key pages returned truncated HTML and pricing was entirely absent. The site is moderately agent-navigable but relies on a machine-readable llms.txt file for substantive content rather than structured HTML; standard pages like /services, /about, and /projects exist but their full text was not extracted by the fetches.

• ›Step [1] (homepage) and steps [6–8] (services, about, projects, contact) all returned 200s but with truncated HTML responses (<4400 chars), meaning the agent could not extract full page content and had to rely on prior knowledge and the llms.txt fallback.
• ›Step [10] (/llms.txt) was the only content-bearing fetch that returned substantial, complete, machine-readable text (80+ projects, founding year, team size, service descriptions, portfolio examples). This single artifact carried the weight of the entire answer.
• ›The site does not publish pricing on any page (steps [2–3] attempted /pricing and /features routes; both returned 404s). The agent acknowledged this explicitly as a gap and correctly noted it as typical for project-based production work.
• ›The site is in Vietnamese; the agent did not note language as a barrier to understanding but it may limit international discoverability.
• ›Step [4] (sitemap.xml) was navigation-bearing but not content-bearing; it revealed the site structure and led the agent to other pages.