The agent assembled a workable overview of Giant Swarm's platform and positioning from the homepage and docs overview, but encountered incomplete pricing disclosure, unreconciled financial claims, and an absence of competitive framing on the site itself. Key details (delivery models, customer names, TCO example) came from the homepage; structural information came from docs; the agent had to flag math errors and fill in component names from inferred navigation labels.

• ›Step [0] (homepage) was the primary content source, yielding delivery models, named customers, TCO example, and differentiation claims. Step [1] (docs root) was a dead-end router; Step [2] (overview) added component and infrastructure detail. The site's main value proposition required piecing together homepage + overview sections.
• ›Pricing is deliberately quote-only and not machine-readable. The TCO calculator example (€440K savings) contains unverified assumptions and an arithmetic error (€480K − €180K ≠ €440K), which the agent flagged but could not resolve from the site itself.
• ›The agent correctly inferred from navigation labels that the platform uses Cluster API, Helm, FluxCD, Cilium, Dex, and Kueue/KubeRay, but the site's docs never name these components in prose. This required model inference from UI labels alone.
• ›The site has no comparison page or positioning against EKS, GKE, AKS, Rancher, or OpenShift. The only competitive framing is 'DIY vs. Giant Swarm.' The agent could not cite any third-party or on-site competitive analysis.
• ›Customer metrics (150+ clusters, 10+ years, €2.5M vs. €1M+ savings) are self-reported on the homepage with no independent verification, audit, or context. The agent flagged these as unverified claims without additional research capability.