The agent successfully compiled a comprehensive overview of Apollo.io's purpose, target audience, pricing model, and competitive positioning by relying almost entirely on third-party review sites and web search results rather than Apollo.io's own website. Apollo.io's website itself (steps 1, 2, 3, 6, 7, 14) returned only HTML shells from a client-side rendered Next.js app with no extractable content, forcing the agent to pivot to external sources for substantive information. The site proved difficult to parse directly, and critical details—especially the confusing credit-based pricing structure and data quality caveats—were only available through competitor review platforms.

• ›Apollo.io's own pages (/, /pricing, /about, /product, /why-apollo, /blog) all returned raw HTML with no readable content due to client-side rendering. The agent could not extract answers directly from the domain.
• ›All substantive content came from external sources cited in steps 8–13: SalesHive, UpLead, La Growth Machine, and comparison platforms. These sources provided feature lists, pricing breakdowns (including the non-obvious credit system), competitor comparisons, and G2 ratings that were absent or inaccessible on apollo.io itself.
• ›The pricing model—particularly the credit system, expiration mechanics, and actual cost of ownership—was identified as confusing and not transparent on Apollo.io's site; this insight came from third-party pricing analyses (step 9, cited sources).
• ›The agent had to rely on 78% prior knowledge plus web search (22%) because Apollo.io's site content was not machine-readable or accessible via direct fetch. No structured data (schema.org, llms.txt, or API-exposed metadata) was leveraged.
• ›The final response explicitly calls out what was hard to find: website navigation clarity, pricing transparency, data accuracy claims, and feature tier breakdowns—all discoverable only through workaround sources, not the official site.