The agent successfully assembled a comprehensive explanation of GoodData.AI's purpose, audience, pricing model, and competitive positioning by combining fetched homepage and product pages with web search results and third-party pricing analysis. The site itself was only partially navigable — the homepage and /product, /pricing, and /company pages loaded successfully, but most predictable feature/use-case URLs returned 404s, forcing the agent to rely heavily on web search and external sources (G2, PricingSaaS, UseDataBrain) to fill gaps on pricing specifics and competitor comparisons.

• ›Step [1] (homepage) provided the tagline and overall positioning but was JavaScript-heavy and lacked detailed feature content in its HTML. Steps [3] (pricing page) and [5] (product page) confirmed key concepts like 'agentic analytics' and 'embedded decision-making' but again without transparent pricing numbers or granular feature lists.
• ›The agent discovered that gooddata.ai has a sparse information architecture: /about, /features, /use-cases, and /how-it-works all returned 404s (steps [4], [7], [8], [9]), indicating either a lean site structure or dynamic routing. Step [15] (/company) succeeded, providing company-level positioning, but predictable resource URLs were missing.
• ›The agent compensated by pivoting to web search (steps [11], [12], [13], [16], [18], [19]), which surfaced third-party pricing aggregators, competitor comparison matrices, and case study links. This external data became the backbone of the pricing and differentiation sections—the site itself did not publish transparent pricing or detailed competitive analysis in a fetchable form.
• ›The final answer explicitly cited sources, including the successful fetches ([1], [3], [5], [15]) and unpublished URLs from search results (e.g., G2, UseDataBrain, Atlan blog). This indicates the agent treated both on-site content and third-party indexing as equally authoritative, which was necessary because the on-site content was incomplete.