The agent successfully understood noticed.so's core positioning, pricing, and differentiation by assembling information from the homepage, pricing page, blog RSS feed, and blog posts. However, the site's heavy JavaScript rendering made direct page scraping uninformative; the agent had to rely heavily on blog content and prior knowledge (38% of fetches) to construct a coherent answer. The site is poorly navigable for agents—expected pages like /features, /how-it-works, and /about returned 404s, and the main pages rendered as JavaScript shells without readable text content.

• ›Steps [1], [3], [8], [14] returned HTML with no visible text content—all pages were JavaScript-heavy Next.js renders with only meta tags and script references. The agent could not extract marketing copy, features, or value propositions directly from these fetches.
• ›Steps [27], [29], [30] were the only content-bearing responses: the RSS feed provided blog post titles and links (revealing engineering focus on 'identity matching' and 'trust'), and the blog post fetches themselves contained substantive information about the product's philosophy and positioning.
• ›The agent had to discover core information (what the product does, who it's for, pricing) through inference from sparse clues: blog post titles, the /speedrun subdomain's purpose (discovered via site search [20]), and prior knowledge of the market. The pricing page fetch [3] returned no readable pricing details despite a 200 response.
• ›Step [24] (speedrun.noticed.so) returned another JS shell, confirming that marketing content is not agent-readable across the entire domain.
• ›The site published a valid RSS feed [27], which was the single most useful structured data for discovery, but blog post content itself remained behind JavaScript rendering and had to be inferred from titles and metadata.
• ›Many expected navigation routes failed (steps [4], [5], [11], [15] returned 404s), forcing the agent to fall back on web search and prior knowledge to fill critical gaps.