The agent successfully understood Clerk's core offering, pricing, and market position, but had to supplement the site's own content with external comparison articles to answer the competitive differentiation question—Clerk's own site does not publish comparisons.

• ›Steps [2] and [3] (homepage and pricing page) were discoverable via direct URL typing, not surfaced through site navigation. The agent guessed both URLs from memory, indicating Clerk's IA may lack clear pricing/feature navigation from the homepage.
• ›Step [2] (homepage) returned high-level positioning but lacked specific competitive claims; step [3] (pricing) gave detailed tier structure and volume-based overages, providing complete pricing data but no usage calculator or cost-at-scale examples.
• ›Step [4] (docs page) was a dead-end for competitive comparison—it returned feature documentation but explicitly lacked comparison content, forcing the agent to web search.
• ›Step [6] (web search) was the only source of substantive competitive differentiation claims (DX, organizations built-in, pricing vs. Auth0/Supabase, use-case logic), meaning Clerk's own site does not answer the 'how it's different' part of the task.
• ›The agent correctly cited all three steps but did not flag that step [2] was behind a domain redirect (clerk.dev → clerk.com), which is a notable findability friction point not mentioned in the final response.