The agent assembled a comprehensive overview of NavGujarat Academy by combining shallow HTML fetches from the marketing site with deep API exploration of the backend service. The marketing site (navgujaratacademy.online) returned only boilerplate meta tags across all pages, forcing the agent to pivot to the VertexPortal API (vertexportal.onrender.com) where actual course, category, and pricing data lived. The site proved only moderately agent-navigable: the frontend was content-hollow, but the backend API exposed structured course and category data that enabled the agent to answer what the platform does, who it serves, and how much it costs.

• ›Steps [1–4] fetched navgujaratacademy.online but returned identical boilerplate HTML with truncated meta descriptions; none of the pages (/pricing, /courses, /about) returned differentiated content, indicating poor site structure or client-side rendering that hid content from server responses.
• ›The agent discovered the VertexPortal backend API (vertexportal.onrender.com) through prior knowledge after exhausting the frontend, not through navigation links or documentation on the marketing site—a critical missing bridge between discovery and implementation.
• ›Step [17] and [26] successfully returned detailed course data (MERN bootcamp title, subtitle, description, pricing: ₹499) and step [21] returned 8 course categories with descriptions—these API endpoints were the only content-bearing sources that answered pricing, scope, and positioning questions.
• ›The agent had to infer the disconnect between claimed content (8 categories, features, live classes, certificates) and reality (1 published course, 8 students) by correlating API pagination metadata with marketing claims—no explicit documentation surfaced this gap.
• ›Steps [18–20, 22, 24–25, 28–29] probed for /pricing, /subscriptions, /instructors, /modules, /testimonials endpoints and all returned 404, showing the backend API is minimally built and lacks structured endpoints for pricing tiers, instructor bios, or module details.
• ›The agent never found instructor qualifications, subscription models, support documentation, class schedules, or how the 'AI tutor' actually works—all cited in marketing but absent from both frontend and API.