grayswan.ai
success
Claude Code
intent

What does grayswan.ai do and who is it for? Explain it back to me.

17steps
77.4sduration
$0.8059cost
172,481tokens
23 steps6 reasoning steps4 searches
home
docs
/product
/features
sitemap
docs
/solutions/platform/cy…
/solutions/platform/sh…
/use-cases/prevent-ai-…
/solutions/ai-red-team…
search
docs
search
/request-demo
search
docs
search
92%
on-site discovery
77%
reliability
85%
link following
path origin
  • previous resource85%
  • web search8%
  • prior knowledge8%
insight

The agent successfully gathered enough information to explain what Gray Swan does, who it's for, and how it differentiates itself, but encountered a significant gap: the site does not publish pricing information, forcing the agent to infer a custom enterprise model. The site was navigable once the agent discovered the sitemap; however, standard navigation paths (/pricing, /product, /features) returned 404s, requiring workaround discovery.

  • ›The homepage (step [1]) and sitemap (step [6]) provided the primary navigation scaffold; the sitemap was essential because conventional URL guesses failed, indicating the site structure is non-standard or minimal.
  • ›Content-bearing steps (about, Cygnal, Shade, use-cases, red-teaming, blog, demo request) all returned 200s and contained substantive product descriptions, team credentials, and positioning language, but none contained pricing—suggesting intentional omission rather than site error.
  • ›The agent had to reconstruct the founding team and funding context from web search results (steps [14], [18], [21]) because the fetched pages themselves did not contain this differentiator explicitly; the 'About' page likely mentioned it, but the agent built a richer narrative by triangulating search results with page content.
  • ›No technical documentation, integration examples, or self-service trial pathway was discoverable on the main site; the 'request demo' page (step [20]) and docs subdomain (step [12]) suggest all customer onboarding is sales-gated, reducing machine-readiness.
  • ›The agent correctly identified and cited sources in its final response, grounding claims in retrieved content (Cygnal runtime protection from step [10], Shade red-teaming from step [11], data breach prevention from step [15]); however, pricing and competitor positioning required inference or external search results rather than on-site content.

Help the next agent finish on grayswan.ai

Front builds the agent version of your site and shows which agents visit, what they came for, and what your pages couldn't answer.

or talk to us about agent readiness →