The agent successfully assembled a comprehensive overview of Insite Teams' purpose, target audience, and positioning by combining web searches with direct fetches from the main insite.co.uk domain and its subpages. However, the site's actual content proved largely inaccessible—all fetches returned truncated HTML dominated by boilerplate (Google Tag Manager, IE compatibility scripts) with substantive page content stripped out, forcing the agent to rely heavily on web search snippets and prior knowledge (40% prior knowledge, 40% web search) rather than direct content extraction. The agent synthesized information from Capterra, GetApp, and QIC Management Systems reviews mentioned in search results, but could not access those review pages directly due to Cloudflare blocks, leaving pricing details incomplete and competitive positioning partly inferred rather than documented.

• ›All five direct fetches from insite.co.uk domains (steps 5, 8, 10, 22, 31) returned responses truncated to ~4400 characters, cutting off substantive content after boilerplate scripts. The agent could not extract actual feature descriptions, pricing tables, or differentiation statements from the primary domain.
• ›The agent discovered that specific pricing tiers (Starter, Premium, Premium Plus) exist but actual prices are not publicly listed—this gap in the final response ('Exact Pricing Tiers: The pricing page metadata suggests...but actual prices aren't publicly listed') reflects the site's intentional opacity rather than agent failure, but makes the site poorly navigable for discovery-evaluation tasks.
• ›Search results (steps 7, 11, 13, 14, 19, 20, 23, 25, 26, 28, 29, 32) were content-bearing for competitive positioning and features (Pulse AI voice assistant, offline capability, snag/punch list focus), but the agent had to assemble these from search snippet text rather than from the primary domain's own documentation.
• ›The agent cited only 2 direct fetches in its final Sources list (insite.co.uk/features and insite.co.uk/pricing) despite fetching 5 URLs, indicating those pages provided minimal usable content. The remaining citations rely on URLs discovered via search, not the primary website.
• ›The final response contains specific claims about features (Pulse AI, offline-first design, voice recording, automatic sync) and competitive differentiation that appear to come from search result snippets and prior knowledge rather than from readable insite.co.uk content, reducing traceability and confidence.