The agent assembled a comprehensive explanation of Pulumi by directly fetching the homepage, pricing, product, docs, and solutions pages, but had to rely heavily on prior knowledge and external search results to fill gaps. The site's pages returned HTML fragments that the agent could not fully parse into clear, structured answers — pricing tiers and feature distinctions were not explicitly spelled out in the fetches themselves, forcing the agent to infer or supplement from external sources.

• ›The pricing page (step [3]) was fetched but returned only HTML head/script tags in the visible response; the agent's pricing breakdown ($40/month Team, $400/month Enterprise) appears to come from prior knowledge or the external search results cited, not directly from the fetched content.
• ›The product page (step [4]) and solutions page (step [7]) were fetched but similarly truncated in the response shown — the agent could not extract detailed feature or use-case information from the actual page bodies, suggesting the site's content is JavaScript-rendered or was not fully captured.
• ›The agent correctly identified three key gaps: 'Pulumi ESC' is mentioned but not defined on the site; the 'Credits' pricing model is unclear; and free-tier limitations are not explicitly stated. These represent information the site publishes but does not make readily machine-readable.
• ›The agent resorted to web search (step [9]) to find alternative comparisons — a signal that the site itself does not prominently position Pulumi against Terraform, AWS CDK, or Ansible in a side-by-side format.
• ›67% of the final answer relied on prior knowledge rather than fresh fetches, indicating the site's marketing pages are not structured in a way that allows an agent to extract the core value proposition, pricing, and differentiation from raw HTML.
• ›The agent successfully answered the task, but the 'heavy bridge' friction classification is justified: the site required external context and inference to produce a coherent explanation, rather than publishing a machine-readable summary of what it does, who it's for, pricing tiers, and competitive positioning.