The agent assembled a comprehensive overview of The Context Company by combining prior knowledge (60%) with fragments fetched from the homepage, pricing, docs, features, and compare pages — but encountered significant friction because the site relies heavily on client-side rendering, making pricing details and comparison matrices inaccessible to the crawler. The agent delivered a satisfying explanation of what the company does, who it's for, and how it differs from competitors, but explicitly flagged that specific pricing amounts and feature-tier boundaries could not be extracted from the website itself.

• ›Steps [1], [3], [5], and [8] returned only HTML shells (truncated at 4400 chars) with no content visible in the raw response — the site uses client-side JavaScript rendering (Next.js with Vercel AI / WorkOS integration), so critical UI content (pricing numbers, comparison tables, feature details) never made it into the fetched markup.
• ›The agent relied heavily on prior knowledge and external search results (web_search: 20%, prior_knowledge: 60%) to construct the final answer because the site's own pages were not machine-readable; the /pricing, /features, /compare, and /docs URLs were guessed or followed as links, but none returned meaningful structured data.
• ›Step [4] (/docs) and step [5] (/features) both returned truncated HTML with no visible feature list or documentation content in the response body, indicating dynamic loading. Step [8] (/compare) similarly returned an empty shell.
• ›The agent explicitly cited external sources (infrabase.ai, MLflow, and the site's own /compare and /knowledge-base URLs) in the final response, indicating it had to work around the site's rendering limitation by pulling information from third-party comparison sites and guides rather than the primary source.
• ›The final answer includes multiple callouts ('⚠️ Not clearly stated', 'unclear', 'didn't load', 'wasn't visible') about missing information — pricing amounts, free tier limits, 'TCC Agent' explanation, data retention limits, comparison matrix details, and enterprise security features — all of which the site fails to expose in a crawlable form.