The agent successfully assembled a comprehensive explanation of Hermoso by fetching machine-readable documentation (llms.txt, pricing.md, index.md, mcp/index.md) and the agency-focused landing page directly from the site. The site publishes structured, agent-friendly content via markdown and meta tags, making discovery straightforward; however, some implementation details (exact per-render costs, video production specs, real-world adoption) remain opaque and required the agent to flag gaps rather than answer them.

• ›Steps [1], [2], [3], [4] returned machine-readable markdown twins (.md files) of the homepage, pricing page, and MCP docs—Hermoso deliberately publishes structured, fetchable versions of its key pages, which the agent parsed directly into its answer rather than scraping HTML.
• ›Step [0] (homepage) and step [9] (agency page) provided HTML with clear meta descriptions and title tags that scaffolded the agent's understanding of what Hermoso does and who it serves, but lacked granular pricing or technical integration details.
• ›The agent correctly identified that credit costs, video model output specs, and campaign performance reporting are not explicitly published upfront—the site defers to 'shown in the app', forcing the agent to work around missing information rather than retrieve it.
• ›Web searches [5], [6] yielded competitor context but did not materially change the answer; they mostly confirmed Hermoso's positioning implicitly by showing alternative tools (AdCreative.ai, Lapis, Creatify). The agent did not cite these search results, suggesting they were used for validation rather than primary sourcing.
• ›The site is highly agent-navigable for what IS published (metadata, llms.txt, markdown docs, clear OAuth endpoint) but deliberately gates operational details (per-render pricing, performance dashboards, case studies) behind the web app, creating predictable gaps in agent-native understanding.