The agent assembled a reasonably complete answer about Rettungshelden's business model, target audience, and competitive positioning, but had to rely heavily on prior knowledge and web search rather than the site itself. The website's JavaScript-heavy architecture made direct content extraction difficult; the agent could not access key pages (pricing, about, course details) through standard fetches, and instead reconstructed the picture from search results, cited Trustpilot reviews, and inferences about German first aid course market conventions.

• ›Steps [9], [10], [15], [18], [19] returned web search snippets that confirmed Rettungshelden operates first aid courses, integrates eye tests and passport photos, and competes against M-A-U-S and Primeros. The agent cited these search results as sources but the actual content came from indexing/metadata rather than fetching Rettungshelden's own pages.
• ›Direct fetches of rettungshelden.com/ and /kurse [steps 1, 3, 12] returned HTML stubs with no readable content—the site is built on Next.js and requires client-side JavaScript rendering to display course listings, pricing, and service descriptions. The agent could not access /preise, /pricing, or /ueber-uns (all 404s [steps 5, 6, 7]), forcing reliance on search results and prior knowledge of German driver's license regulations.
• ›The agent's answer includes specific pricing (€39.99–€55.00), competitive Trustpilot ratings, and service bundles (Sehtest + Passbild) that were NOT found on rettungshelden.com itself but inferred from search results and market knowledge. The site surfaced none of this information in a machine-readable form; it is fundamentally not agent-friendly for structured discovery.
• ›The agent explicitly cited Trustpilot reviews as a source but could not fetch that page (403 [step 13]); it relied on search result metadata instead. This suggests the agent's final answer about ratings and competitor comparisons came from prior knowledge + search indexing, not live retrieval.