The agent successfully gathered enough information to explain Rensei's core purpose, target audience, and differentiation, but encountered significant gaps in pricing transparency and competitive positioning. The site's heavy use of compressed Next.js with minimal accessible text forced the agent to rely heavily on web search and prior knowledge (83% of fetches came from search or prior knowledge) rather than directly readable site content, making it difficult to extract complete information natively.

• ›The homepage, pricing page, and about page returned heavily minified Next.js HTML with no extractable text content—the agent could not read the actual value proposition or pricing from the primary sources it visited ([1], [3], [4]).
• ›The agent discovered Rensei's core differentiator (deterministic runtime, audit trails, Cedar policy integration) primarily through web search results and the GitHub repository ([7], [10], [11]), not from the website itself. The search result snippet from [7] contained the tagline 'The deterministic runtime that makes agent-authored code defensible in production,' which became the foundation of the explanation.
• ›Pricing information was explicitly inaccessible—the agent found only marketing language ('Pricing is scoped to each deployment') rather than actual tier details or dollar amounts. The agent correctly identified Donmai as open-source/free and Rensei as custom enterprise pricing, but this required inference and search rather than published site content.
• ›The agent relied on the GitHub Donmai repository ([11]) and donmai.dev ([12]) to understand the technical architecture and open-core model, as the main Rensei site did not surface these details in a readable format.
• ›Competitive positioning was never directly addressed on the site—the agent had to assemble this from search results about Cedar policy adoption at AWS and general agent platform comparisons, not from Rensei's own 'vs. alternatives' content.
• ›The site is not agent-readable in its native form; all substantive content extraction required web search augmentation or external sources (GitHub, LinkedIn). This indicates poor semantic structure, missing schema markup, or intentional client-side rendering that obscures content from web crawlers.