The agent successfully assembled a comprehensive overview of bilder.ntb.no by combining fragments from the site's HTML metadata, NTB's corporate pages, and external documentation. The site itself is poorly navigable for machine agents: the primary bilder.ntb.no domain serves only a React shell with minimal indexable content, forcing reliance on web search to discover what NTB actually offers, who uses it, and how it differs from competitors. Pricing information remains completely inaccessible and was correctly flagged as unavailable.

• ›Step [1] only revealed a meta description (300M images/videos) and React boilerplate—the core content is client-rendered and not available in initial HTML. Steps [4] and [13] to bilder.ntb.no/en and bilder.ntb.no/r returned identical React shells, confirming the site does not surface navigation or feature information server-side.
• ›The agent discovered what NTB actually does by fetching external resources: ntb.no/bilder (corporate landing page, step [28]), ntb.no/bilder/redaksjonelle-foto-og-videooppdrag (editorial services page, step [30]), and mediebank.ntb.no/p/about (product documentation, step [19])—all outside the bilder.ntb.no domain. This indicates NTB's main website does the positioning work; bilder.ntb.no itself is a portal, not a marketing or information site.
• ›Pricing was never found anywhere on the site or in official documentation. The agent correctly concluded it requires direct sales contact, but this is a significant gap in agent-readiness for a B2B SaaS-like service. No public pricing page, tier list, or API documentation on pricing exists.
• ›The agent struggled to distinguish between bilder.ntb.no, NTB Mediabank, and NTB Lens. These are related services (archive portal, DAM platform, mobile app) but their relationships and use cases are not clearly explained on any single resource or official documentation. The agent had to infer this from scattered fragments.
• ›Web search (67% of fetch sources) was essential to success. The agent could not have completed the task using only bilder.ntb.no or even ntb.no; external Wikipedia entries, corporate profiles, and blog posts filled critical gaps. This suggests the site itself is not designed for discovery or self-serve understanding.