The agent worked around a Cloudflare-protected website to assemble a comprehensive explanation of Caivan by relying entirely on web search results and third-party sources. It successfully identified Caivan as an Ontario homebuilder with AI-driven manufacturing, pricing strategies including tax-free offerings, and differentiation through quality assurance and sustainability—but had to piece this together from fragments across multiple search results rather than from direct site content, making the task a discovery-evaluation effort constrained by poor site navigability.

• ›The main domain and all internal pages returned Cloudflare 403 blocks, forcing the agent to bypass direct site access entirely. Every substantive piece of information came from web search snippets and third-party aggregators (ZoomInfo, RENXHomes, CondoTrend, martini.ai, BBB, Ottawa home review sites)—not from caivan.com's own published content.
• ›Search results themselves contained answer fragments: step [2] returned link titles hinting at 'Tax-Free Pricing' and 'Our Team'; step [12] returned the concept but not mechanics; steps [14] and [15] surfaced the ABIC/AI manufacturing angle via news coverage. The agent had to cross-reference multiple searches to build a coherent narrative.
• ›The site publishes key differentiators (Caivan 360 QA program, ABIC facility, tax-free pricing, sustainability) but only as page titles and URLs in search results—not as accessible, machine-readable content. The agent could not fetch the actual 'Tax-Free Pricing,' 'Homeowner Journey,' 'Our Approach,' or 'Caivan 360' pages to verify details or quote directly.
• ›Pricing information was fragmented: the agent found scattered references to price ranges ($290K–$3.4M) but noted 'no clear public pricing guide' and acknowledged that base pricing is not publicly listed. This data came from third-party aggregators (precon.realestate, TrueCondos) rather than Caivan's own price pages.
• ›The agent's final answer explicitly lists what it could NOT find: Homeowner Journey details, specific customer satisfaction metrics, competitive head-to-head analysis, and ABIC technical specs—all due to site inaccessibility or lack of public disclosure, not agent failure.