The agent successfully assembled a comprehensive answer about Cardinal Gray's business, but had to work around a nearly inaccessible main website. The cardinalgray.com homepage and standard product pages (pricing, about) returned only empty Next.js shells with no readable content. The agent relied almost entirely on third-party sources (web search results, CB Insights, LinkedIn, HireTOP blog) to reconstruct the company's value proposition, pricing model, and competitive positioning—none of which were directly surfaced by the site itself.

• ›The main website (cardinalgray.com/) is a dynamic Next.js single-page app that returns HTML but no visible content in the response body. Standard navigation paths (/pricing, /about, /product/lender) all returned 404s, forcing the agent to rely entirely on external sources.
• ›Web search results [7, 8, 13] were content-bearing, providing links to third-party business intelligence platforms (CB Insights, ZoomInfo, Extruct AI) and media articles. The agent pulled most substantive information from these external sources rather than from cardinalgray.com itself.
• ›The docs subdomain (docs.cardinalgray.com) was accessible but appears to be technical/API documentation behind a minimalist interface—not a public product overview. LinkedIn [15] and the HireTOP blog article [14] were the most informative external sources, providing context on the DMV automation problem, use cases, and customer examples.
• ›Cardinal Gray's main site is a poor discovery and identity layer for an AI agent: no structured data, no public pricing, no feature descriptions, and no documentation accessible without authentication. The company relies on a B2B sales-driven model where prospects must request a demo to learn details.
• ›The agent successfully identified key gaps in what the website publishes (transparent pricing, feature documentation, competitor comparisons) and explicitly flagged these as missing rather than fabricating answers.