The agent assembled a comprehensive overview of Lyft's business model, pricing structure, and competitive positioning by combining the homepage and rider/driver landing pages with help documentation and external web search results. The site itself exposed limited pricing detail (the main domain returned 404s for /pricing and /about), forcing the agent to rely heavily on prior knowledge (54%) and help.lyft.com articles discovered through search. The agent successfully answered all four required elements of the task (what they do, who it's for, pricing, differentiation) but had to work around information gaps on the public site.

• ›Steps [1], [5], [6] returned landing pages that explained Lyft's core offering (ride-sharing for riders and drivers) and positioning but lacked pricing details; the homepage meta description and titles confirmed 'upfront pricing' and 'rewards' but no actual rates.
• ›Steps [8], [9], [16] accessed help.lyft.com articles on pricing and ride types, which contained the substantive pricing breakdown (base fare, per-mile rates, service fees, surge pricing mechanics) cited in the final response—but these required direct URL guessing or web search discovery, not navigation from the main site.
• ›Steps [13], [14] performed web searches that surfaced external comparisons and help articles; the agent did not find pricing or competitive differentiation on lyft.com itself and had to rely on third-party sources (ridester.com, whitelabelfox.com, enatega.com) to answer the 'how is it different' question.
• ›The main domain has poor information architecture for agent discovery: /pricing, /about, /bikes-scooters returned 404s; /faq returned 429 rate-limit. The agent had to guess at help.lyft.com subdomains and run web searches to find pricing, ride types, and competitive positioning.
• ›The site is not agent-ready for self-service pricing lookup—users must download the app and enter location/destination. This information is published only on help.lyft.com and in external third-party analyses, not on the public marketing site.
• ›The final response cites sources from both on-domain (help.lyft.com articles) and external sources, indicating the agent had to assemble the answer from fragments across multiple origins, not from a single cohesive published resource.