The agent successfully assembled a comprehensive explanation of Glean by combining limited content from the site itself (homepage, about page, agents guide) with extensive prior knowledge and web search results. The site's JavaScript-heavy architecture and sparse public documentation meant that concrete details about what Glean does, who uses it, pricing, and competitive positioning had to be sourced primarily from external analyses rather than Glean's own pages.

• ›Steps [1], [8], and [13] returned valid HTML but with minimal extractable content detail—the agent relied heavily on prior knowledge (38% of the run's fetch sources) and web search results (13%) to fill critical gaps. The homepage and about page exist but apparently lack the semantic detail needed to answer the task without external context.
• ›The site is JavaScript-dependent and truncated in the fetch responses, making it difficult to extract feature descriptions, customer examples, and pricing rationale directly. The agent had to guess at URLs like /pricing, /product, /why-glean, and /enterprises, most of which returned 404s, forcing a pivot to web search for pricing and competitive data.
• ›Glean deliberately obscures pricing and detailed use-case positioning behind enterprise sales workflows—no public price list or transparent ROI metrics are published on the site. The agent correctly identified this as a confusing gap and supplemented from third-party analyst sites (Vendr, Coworker AI, GoSearch) and Dust blog comparisons, which were far more informative than Glean's own marketing pages.