The agent successfully synthesized an accurate overview of Spearhead's business model, positioning, and service offerings, but had to work around a client-side rendered website that obscured content visibility. Direct fetches of the homepage and key pages (/about, /work, /basecamp, /agentic-enterprise) returned minimal textual information due to JavaScript rendering, forcing reliance on web search to surface concrete details about services, pricing models, and differentiation. The site is poorly agent-readable but the agent recovered by inferring from search results and industry patterns.

• ›Steps [0], [1], and subsequent direct page fetches returned HTML skeletons with truncated content — the website uses heavy client-side rendering, leaving the agent unable to extract service descriptions, team details, or engagement models directly from the DOM.
• ›The agent never retrieved a dedicated pricing page (step [3] returned 404), and no web search result contained Spearhead's actual price point, forcing the agent to disclose the gap and estimate based on industry benchmarks rather than citing primary sources.
• ›Web searches (steps [2], [10], [12], [16], [17], [18], [22]) returned metadata and snippet fragments that proved more informative than the site itself — the agent reconstructed 'pod-based delivery,' '90-day engagement,' and service positioning from search result titles and descriptions, not from on-site content.
• ›The agent explicitly cited six pages in its Sources section ([0], [1], [4], [5], [7], [13]), but only [4] (/about) and [5] (/work) were likely to have returned meaningful rendered content; the others appear cited for completeness rather than because they directly answered questions.
• ›AI Basecamp (step [13]) was the only page that successfully surfaced a concrete, differentiating product offering (the 7-stage free learning program), demonstrating the site can expose value when the agent reaches the right URL.
• ›The NVIDIA partnership, pod model, and 90-day engagement timeline were all confirmed via web search (steps [17], [16]) rather than from direct site navigation, indicating the site underexplains its core value props in publicly crawlable form.