waniwani.ai
success
Claude Code
intent

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

16steps
63.3sduration
$0.8240cost
205,267tokens
20 steps4 reasoning steps5 searches
home
docs
docs
/features
/partners
/solutions
/how-it-works
search
/2026/06/16/waniwani-r…
search
/views/waniwani-raises…
search
docs
search
search
/compare/marrow-vs-wan…
64%
on-site discovery
64%
reliability
27%
link following
path origin
  • previous resource27%
  • web search36%
  • prior knowledge36%
insight

The agent gathered enough information to provide a usable explanation of Waniwani's business, but had to work around significant gaps in the site's own documentation. The homepage and public site pages (about, partners, blog) contained only high-level positioning; the agent relied heavily on external sources (Seedcamp announcement, FinTech Global, news articles) and prior knowledge to assemble a complete picture. Critical information like pricing was not published anywhere the agent could access, forcing it to acknowledge uncertainty and note what was missing.

  • ›Steps [1], [4], [5], [18] (waniwani.ai pages) returned Next.js-rendered HTML that was truncated in the fetch response, making it impossible to extract textual content directly from the site. The agent could not read the actual homepage, about page, or blog post content.
  • ›Steps [14], [15] (external news articles from Seedcamp and FinTech Global) were content-bearing and explicitly cited. These third-party sources provided the core facts about funding, business model, target customers, and competitive positioning—information that should have been on Waniwani's own site but was either not present or not accessible.
  • ›Step [9], [10], [17] (web searches) yielded URLs and snippets that pointed the agent toward external coverage; the searches themselves didn't contain full answers but enabled discovery of Seedcamp and news articles.
  • ›The site's information architecture is sparse—expected routes like /pricing, /features, /solutions, /how-it-works all returned 404s, signaling either incomplete navigation or a deliberately minimal surface area. The agent had to guess at what existed.
  • ›Pricing is entirely absent from publicly accessible resources. The agent correctly flagged this as a major gap and inferred a likely 'freemium SDK + paid infrastructure' model based on context clues, but could not confirm tier structure or cost.
  • ›The agent's final response is grounded and honest about what it found versus what it could not verify, citing external sources explicitly and calling out confusion. However, this required heavy reliance on third-party coverage to compensate for thin site documentation.

Want to run your own?

Join the waitlist for early access to point your own agents at any domain, with the intents you choose.

or talk to us about agent readiness →