The agent assembled a coherent explanation of onRun by relying almost entirely on app store listings and a Tracxn company profile, since the onrun.com website itself proved nearly opaque—homepage and subpages returned either truncated HTML or 404s. The agent successfully identified what onRun does (race-focused running social app), who it's for (community-oriented runners), pricing (free), and key differentiators (automatic race import, emphasis on community over competition vs. Strava). However, critical details like race data sources and actual feature parity remained uncertain, and the site's own marketing pages were inaccessible.

• ›Steps [6] (Apple App Store) and [11] (Google Play) returned content-bearing responses that provided the core value prop ('Race, Pace, Chase'), user intent signals, and download availability—this was the primary source of what onRun *is*.
• ›Step [14] (Tracxn company profile) was cited in sources and likely contained founder name, mission hints, and competitive positioning—filling gaps the official website could not.
• ›The onrun.com homepage ([0], [19]) and /about page ([3]) returned only truncated Next.js HTML with no visible text content extracted; /pricing and /features returned 404s ([1], [2]). The agent had to infer pricing (free) from app store language and context, not from an official pricing page.
• ›The agent relied on 14 web searches (steps [4]–[5], [7]–[10], [12]–[13], [15]–[18], [20]–[24]) to backfill missing information, suggesting the website itself was not serving as the primary information surface—discovery and positioning were delegated to third-party platforms and search.
• ›The agent explicitly flagged three confusions: which race databases auto-populate, whether premium features exist despite the 404 pricing page, and lack of Reddit/mainstream visibility—all indicating incomplete or contradictory signals from available sources.