The agent retrieved the core facts needed to explain the site (what, who, pricing) from three public pages, but could not answer how it differs from competitors because the site itself makes no such comparison. The pricing page was authoritative and complete, though billing models (subscription vs. one-time) and refund terms remain undocumented on fetched pages.

• ›The homepage and pricing page were discovered by memory (typed URL), not surfaced by search or site navigation—the agent had to guess the URL structure.
• ›Pricing was fully and clearly listed on /cennik with VAT transparency, but the page never discloses whether courses are subscriptions or one-time purchases, and refund terms are gated behind a /regulamin link that was not fetched.
• ›The about page disclosed company registration details and historical founding narrative but made only unsubstantiated claims about results (64k students, 98% recommendation rate, 80% pass rate) with no methodology or sample sizes disclosed.
• ›The site does not compare itself to competitors or alternatives—no differentiation is claimed beyond feature recitation. The agent correctly noted this as model memory gap, not a site deficiency.
• ›Course naming and edition labeling (mixing 2024, 2026, 'do matury 2027') creates ambiguity on what a buyer should purchase; the pricing page aggregates but does not clarify which edition is current.