The agent successfully assembled a comprehensive explanation of BNP Paribas's business model, divisions, market position, and differentiation strategy by combining information from the official group.bnpparibas site structure with supplementary web search results and external sources. However, the site itself—which is JavaScript-heavy and corporate-focused—did not directly expose pricing details, specific product catalogs, or retail customer information; the agent had to rely heavily on external sources (60% web search sourcing) to construct a complete answer, and explicitly noted multiple information gaps that existed on the actual domain.

• ›Fetches [7], [8], [9] from group.bnpparibas returned only HTML scaffolding with meta descriptions; the agent noted these pages are 'JavaScript-heavy and not loading the full content,' meaning the actual substance of those pages (business division details, strategy, governance) was not extractable from the direct fetches alone.
• ›The agent relied on web search results [11], [12], [14], [15], [17] to answer most questions about pricing, competitors, divisions, and differentiation—these external sources (Umbrex, PitchGrade, MBA Skool, InfotechLead) contained the substantive answers, not the official site itself.
• ›The official site explicitly cited in the final response [group.bnpparibas/en/group/our-businesses-and-expertise, group.bnpparibas/en/group/about-us] did NOT deliver usable content via fetch; they were cited because their URLs were topically relevant and listed in search results, but the agent had to obtain actual business division definitions and strategy from external third-party analysis.
• ›The agent correctly identified that group.bnpparibas is a corporate/investor relations hub, not a customer-facing product portal, explaining why no retail pricing, product catalogs, or account-specific details were discoverable—this is a fundamental architectural limitation of the domain.
• ›The site is poorly agent-ready: JavaScript rendering required, no machine-readable structured data returned in fetches, no transparent pricing published, and no comprehensive product/service directory. The agent had to work around these constraints by searching external sources.