The agent successfully gathered enough information to explain State Machines' core offering, target audience, and go-to-market approach, though pricing transparency and technical details remained unavailable. The site is minimally navigable for discovery: the homepage and standard routes (/pricing, /docs, /about) returned 404s, forcing reliance on the llms.txt artifact and inference from /start and /book landing pages. The agent pieced together a coherent narrative from fragmented, non-standard sources rather than from dedicated product pages.

• ›Step [6] (llms.txt) was the primary content source: it directly stated what State Machines does ('local, stateful replicas of enterprise APIs'), who it serves ('teams building and evaluating AI agents'), and which systems it supports ('Salesforce, Zendesk, SAP, Workday, and others'). This is the only machine-readable, structured explanation the site publicly offered.
• ›Steps [8] and [9] (/start and /book) were cited in the final response but returned truncated HTML. The agent inferred their function (intake form, booking) from context and URL naming, not from readable content. No pricing details were discoverable on any page.
• ›The site failed standard navigation discovery: /pricing, /docs, and /about all returned 404s, suggesting either a single-page app with client-side routing, a heavily truncated DOM, or intentional path hiding. The agent could not access typical product documentation or comparison pages.
• ›The agent relied heavily on prior knowledge and the llms.txt artifact (86% of fetches sourced from 'previous_artifact') rather than on published product pages. This indicates State Machines publishes for LLM agents better than for human browsers, yet even that artifact lacked pricing, competitive positioning, and technical integration details.
• ›Pricing is explicitly gated behind a sales conversation (routing to /start and /book), signaling enterprise-only pricing and custom negotiation. This is a legitimate model but adds friction to agent-driven discovery and evaluation.