The agent assembled a comprehensive overview of developer.adobe.com by combining direct site exploration with web search and prior knowledge, ultimately satisfying the task despite significant friction. The site's fragmented structure, missing pricing pages (404s), and scattered documentation required the agent to work around inaccessible resources and piece together information from multiple fetch attempts and external sources. The final answer is substantive and well-sourced, but the agent's reliance on prior knowledge (43%) and web search (14%) reveals that Adobe does not publish key information (especially pricing) in a machine-readable, discoverable way.

• ›The homepage [1] and /apis page [2] provided high-level positioning but lacked specifics; product-specific pages like Firefly Services [6, 17] and Lightroom [11] returned documentation but no pricing details.
• ›Critical pricing information was not on the site at all—/pricing returned 404 [3], forcing the agent to find external third-party pricing estimates via web search [12]. The only internal pricing page found was /document-services/pricing/main [13], which was product-specific and did not generalize.
• ›Documentation is highly fragmented across different URL patterns (/firefly-services/docs/, /lightroom/, /experience-platform-apis/) with no unified entry point—attempted /docs and /getting-started both returned 404s [4, 7], suggesting Adobe intentionally routes developers to product-specific docs rather than a consolidated portal.
• ›Experience Platform APIs [16] was discovered via web search, not from the homepage, indicating poor discoverability of key products from the main site.
• ›The AI Registry [5] page exists but is minimally informative (frame-based, unclear purpose), demonstrating incomplete information architecture.
• ›Competitive positioning had to be inferred from external sources [15] rather than explicit statements on the site; Adobe does not publish 'how we differ from alternatives' content.
• ›The site relies on a sales-driven pricing model for most APIs (no published rates), which is fundamentally opaque to automated discovery and agent evaluation.