The agent successfully assembled a comprehensive explanation of Edge Delta by combining direct site fetches with external sources. The homepage, pricing page, about page, and llms.txt provided core product positioning and pricing structure; the agent supplemented with a third-party pricing review (CubeAPM) to fill in cost details and used web search results to contextualize competitive positioning. The site was moderately navigable—key pages were discoverable and rendered, but JavaScript-heavy product pages and comparison pages returned only HTML shells without readable content, forcing reliance on structured metadata (llms.txt, page titles, meta descriptions) and external sources for deeper differentiation details.

• ›Step [7] (llms.txt) was the most content-bearing resource on the site itself—it contained the clearest plain-text summary of what Edge Delta does (AI Teammates + Telemetry Pipelines), the four specialized teammates, and product architecture. The agent explicitly cited this as a source.
• ›Steps [1, 2, 5, 12] returned HTML pages whose rendered content was inaccessible in the raw response (JavaScript SPAs); the agent extracted only metadata (title tags, meta descriptions) from these pages, not body content. Meta descriptions alone provided enough positioning language ('Root causes, evidence, and timeline ready by the time you do') to anchor the core value prop.
• ›Step [17] (third-party CubeAPM review) was critical for answering the pricing question concretely—the agent's own fetch of /pricing [2] returned HTML without visible pricing numbers in the response. The agent had to leave the site and cross-reference external reviews to provide specific dollar amounts ($20/month entry, ~$9,344/year).
• ›Competitive differentiation was mostly assembled from prior knowledge and web search context [15, 16], not from Edge Delta's own comparison pages. Step [8] (Edge Delta vs Cribl) returned HTML but no readable content; the agent could not extract the actual comparison from the site itself.
• ›Site structure gaps: /features and /how-it-works returned 404s [3, 4], suggesting incomplete navigation architecture. The agent had to work around missing pages and rely on indirect discovery via sitemap and llms.txt.
• ›The agent's inability to extract rendered content from JavaScript-heavy pages created friction—it had to cite external sources (CubeAPM, PR Newswire, Middleware) for pricing tiers, cost breakdowns, and competitive positioning rather than relying on the primary site. This is a 'light bridge' outcome: the agent succeeded despite the site's limited machine-readability for newer product details.