The agent assembled a comprehensive overview of Metriport by combining fragments from the live site (homepage, /about, /medical, /customers pages) with extensive web search and third-party sources. The site itself was poorly navigable for complete self-service discovery: the dedicated pricing page returned 404, several expected navigation routes were broken, and core information (especially pricing tiers and deployment costs) was not publicly documented. The agent worked around these gaps by using search results, GitHub, Y Combinator, and external SaaS databases, ultimately delivering a thorough explanation but only after heavy reliance on external sources rather than the site's own content.

• ›Steps [1], [4], [12], [13], [23] returned valid pages but contained limited structured content—homepage and /medical pages appear to be client-rendered React apps where actionable content was truncated in the HTML response. The agent could not extract detailed feature lists or pricing from these fetches alone.
• ›The pricing page ([2]) returned 404, forcing the agent to search externally. SaaSWorthy ([24] search result, step [18] fetch from external blog) and search snippets provided the only publicly visible pricing data ($0.20/user/month cloud tier); self-hosted pricing was never found on Metriport's own domain.
• ›The agent discovered key content via web search (steps [15], [16], [20], [22], [24], [25], [26], [27]) and external sources (Invene blog, Y Combinator, GitHub, SaaSWorthy) rather than site navigation. These searches yielded more actionable information (open-source positioning, FHIR R4 support, customer names) than direct site fetches.
• ›Docs were hosted at a different domain (docs.metriport.com, step [14]), not at metriport.com/docs, indicating fragmented information architecture that an agent would not discover without prior knowledge or search.
• ›The site's responsiveness to agent discovery was low: multiple standard paths (/pricing, /solutions, /product, /docs, /developers, /why-metriport) returned 404, suggesting either incomplete navigation hierarchy or incomplete site build. The site exposed minimal structured metadata or machine-readable content, forcing the agent to rely on prior knowledge, search results, and external analyses.