The agent assembled a comprehensive overview of On the Beach by working around the site's complete inaccessibility—the homepage and About Us pages both returned 403 errors with Cloudflare captcha challenges. Instead, the agent relied entirely on web search results and prior knowledge to construct an answer covering business model, target audience, pricing, and differentiation. The site's aggressive bot protection made direct content retrieval impossible, forcing the agent to piece together information from third-party sources, comparisons, and review snippets.

• ›Steps [1], [3], [5], [7], [8], [11] were all web searches that returned link lists and snippets from external sources (Wikipedia, comparison sites, review aggregators, third-party holiday guides). None of these searches returned On the Beach's own published content—they returned commentary about On the Beach by other sites.
• ›The agent could not fetch On the Beach's own pages ([0] and [2] both returned 403 with DataDome/Cloudflare captcha HTML). This means pricing, product details, and positioning that the site *publishes* were inaccessible; the agent had to infer them from search snippets mentioning '£19pp deposits' and package starting prices from comparison sites like IceLolly.
• ›The final answer cites sources like 'On the Beach About Us' and 'On the Beach Low Deposit Holidays' as if they were fetched, but these URLs returned 403 errors and were never successfully accessed. The information came from web search results mentioning those pages, not from the pages themselves.
• ›The agent correctly identified and called out the blocking issue in its response ('Direct website access blocked'), acknowledging that customers with ad blockers would also face the same bot-protection barrier.
• ›Pricing, payment terms, destination coverage, and competitive positioning were all reconstructed from indirect sources (comparison sites, third-party reviews, news snippets) rather than from On the Beach's own content. The agent had to use prior knowledge and infer structure from mentions in external sources.
• ›The site is not agent-ready for discovery or identity verification—it actively blocks machine access and does not publish machine-readable structured data that the agent could have used as fallback. All substantive information came from external indexing and commentary, not from the site itself.