The agent successfully assembled a comprehensive overview of Dust.tt, but had to work around a largely client-rendered website that blocked direct content extraction from official pages. Critical information—what Dust does, who it's for, pricing, and differentiation—came primarily from third-party review sites (Coda One), technical deep-dives (ZenML), and industry analysis (Latent Space) rather than Dust's own marketing materials. The official Dust.tt domain returned Next.js shells with no visible HTML content, making the site effectively opaque to automated navigation.

• ›Official Dust.tt pages (/, /pricing, /home/about, /home/pricing, /blog) all returned boilerplate Next.js HTML with Datadog RUM scripts but no extractable content—full page rendering is required to access any marketing copy or pricing details.
• ›Pricing and core product information came entirely from third-party aggregators (Coda One, AISO Tools) and technical analyses (ZenML LLMOps Database, Latent Space), not from dust.tt itself. The agent cited dust.tt/home/pricing as a source, but that fetch [21] returned only client-side boilerplate; the actual pricing data was extracted from codaone.ai [10] and aiso tools.
• ›Documentation site (docs.dust.tt) was accessible and returned renderable HTML [7], confirming Dust maintains a separate technical docs property, but this required web search discovery rather than being linked from the main site.
• ›The agent correctly identified that the pricing page returned a 404 [3], yet was able to infer a credit-based pricing model from aggregator sites—this represents working around missing official information rather than finding it on the domain.
• ›Third-party sources provided consistent positioning language ('horizontal enterprise agent platform', '88% daily active user rates', infrastructure-first approach) that the agent used to build its differentiation narrative, suggesting Dust.tt's own messaging is either not publicly indexed or only visible client-side.