Agent successfully gathered comprehensive information about Édifice's products, target audience, and pricing by navigating the site and using web search to locate pricing pages. The site required deliberate URL discovery (pricing pages were not directly linked on the homepage) but once located, all required facts were available in fetchable form.

• ›Homepage did not surface pricing links directly; agent had to discover pricing URLs via web search before fetching them, indicating poor IA for cost information.
• ›Steps [0], [1], [2] fetched the homepage three times and returned fragmented answers (overview, admission of missing pricing, differentiation); the site front-loaded marketing narrative but hid pricing behind separate URLs.
• ›Steps [8] and [9] successfully retrieved complete, structured pricing tables for both ONE and NEO tiers with per-student-per-year costs, discounts, and feature summaries—site content was machine-readable once located.
• ›Agent's gap analysis ('Ultima tier vague', 'no competitor positioning', 'contact-to-quote model') derived from model reasoning about absent content, not from site contradictions or failed fetches; site simply did not address these points.
• ›Language variation: agent accessed English version (edifice.io/en/) but site is primarily French; no indication that switching language yielded different or more complete answers.