The agent successfully assembled a comprehensive explanation of Ardent's offering, pricing, and differentiation by combining the homepage and pricing page fetches with supplementary research from Y Combinator, third-party review sites, and web search. However, the tryardent.com site itself proved minimally helpful—the agent retrieved HTML from the homepage and pricing page but could not extract rendered content from them (both returned truncated Next.js bundles), forcing reliance on external sources (60% web search, 20% prior knowledge) to construct the answer.

• ›Steps [1] and [2] (homepage and pricing page) returned Next.js-rendered HTML but the responses were truncated and did not expose readable content; the agent could not extract pricing tiers, feature details, or positioning from the primary site itself.
• ›Steps [8] (Y Combinator), [14] (aicoolies.com), and [18] (tools.stoa.agency) were the primary sources of factual content used in the final answer—all third-party sites, not tryardent.com. The agent cited Y Combinator, aicoolies, and Stoa in its sources section.
• ›The site's own /features and /about routes returned 404s (steps [3], [4]), further limiting direct discovery of product positioning; the agent had to reconstruct the feature set and audience from external sources and prior knowledge.
• ›Pricing information in the final response (Community free tier, Scale at $250/month, per-CU-hour and per-GB costs) came from web search results and third-party reviews, not directly parsed from the tryardent.com/pricing page.
• ›The agent identified multiple gaps it could not fill (free tier quotas, specific feature matrix, compliance certifications, data residency) that would likely be documented on the official site if rendering were accessible, indicating the site has relevant information but it was not retrievable in this run.