ora
thelephant.io
success
Claude Code
intent

What does thelephant.io do and who is it for? Explain it back to me.

8steps
54.4sduration
$0.4545cost
85,274tokens
13 steps5 reasoning steps3 searches
home
docs
docs
search
/company
search
/organization/privateq…
search
60%
on-site discovery
60%
reliability
40%
link following
path origin
  • previous resource40%
  • web search40%
  • prior knowledge20%
insight

The agent assembled a mostly complete answer to the discovery-evaluation task by combining fragments from the site itself (homepage, about, company pages) with web search results and prior knowledge, but encountered significant gaps in pricing transparency and technical mechanics. The site publishes its value proposition and use cases adequately, but lacks machine-readable pricing data, detailed feature documentation, and public competitor comparisons, forcing the agent to acknowledge what could not be found rather than provide incomplete guesses.

  • Step [1] (homepage) and steps [4], [7] (about/company pages) returned basic positioning and history (founded 2015, serves 100+ companies, 2,400+ investors) but were heavily JavaScript-dependent and incomplete when fetched in raw form. The agent had to infer much of this from search snippets in steps [6], [8], [11] rather than direct page content.
  • Pricing information was completely absent from the site. Step [3] returned a 404 at /pricing, and none of the subsequent fetches ([1], [4], [7]) contained fee schedules, transaction costs, or tier structures. The agent correctly identified and flagged this as a critical gap rather than fabricating details.
  • The site is moderately agent-hostile: it relies on heavy JavaScript rendering (as the agent noted), does not publish structured data or machine-readable pricing, and does not surface direct competitor comparisons. The agent had to use web search (steps [6], [8], [11]) to piece together what The Elephant does, who it competes with (Forge Global), and its market position—information that should have been on-site.
  • Steps [6], [8], [11] (web search) proved more reliable than direct site fetches for extracting business model details. Search snippets explicitly mentioned 'secondary market platform,' 'late-stage private tech,' and 'Israeli startup' context, while the raw HTML fetches did not render sufficiently to expose the same clarity.
  • The agent successfully disambiguated the three user personas (companies, institutional investors, shareholders) and the core value proposition (controlled liquidity for pre-IPO shares) from disparate page fragments and search results, demonstrating good synthesis but also exposing the site's lack of a clear, centralized pitch.

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 →