Agent successfully synthesized a complete product overview from the site's actual.ai domain (homepage, pricing, docs, and about pages), then augmented it with web search context on competitors. The site was reasonably navigable once the agent bypassed the app.actual.ai app dashboard and found the marketing/documentation site at actual.ai, where pricing and feature content were directly accessible and clear.

• ›The site publishes pricing and core positioning clearly on dedicated pages (/pricing, /about, /docs), making it agent-friendly for this discovery task.
• ›The app.actual.ai domain was a deployed dashboard with minimal content (logo only); the actual marketing site at actual.ai carried all substantive information. This dual-domain structure created initial confusion but was resolvable via direct URL guessing.
• ›Feature differentiation between pricing tiers was not explicitly spelled out on the pricing page itself—the agent had to infer tier value from fragment mentions ('Architecture Agent MCP', 'Management Agent') rather than a feature matrix or tier comparison table.
• ›The site does not directly name or compare itself against known competitors (Guardrails AI, NeMo, Amazon Bedrock Guardrails). The agent filled this gap via web search, not site content.
• ›Integration points (CLI, GitHub, Slack, MCP, and specific AI tool support for Claude Code/Cursor/Copilot) were scattered across docs and the about page rather than consolidated in a single 'integrations' or 'how it works' section.