StartupWiki Launches as a Free, No-Account Alternative to Crunchbase

A new startup research directory went live on Hacker News this week, offering free access to company profiles, search, and filtering without requiring an account or subscription.

A Free Database Built Without Logins or Paywalls

The creator behind the launch post described building the site out of frustration with how hard it can be to research early-stage companies on existing platforms, which often require accounts or charge for access. The goal, in the creator's words, was a site that works like a reference page: no logins, no subscriptions, no opaque scoring systems. The current version includes company profiles, search and filtering, and basic categorization, with a public API still in development. The site's own description positions it as a community-driven, AI-powered directory spanning sectors including AI, biotech, cleantech, and fintech.

AI Agents Build the Profiles, and Some Early Errors Have Surfaced

Behind the scenes, the creator runs AI agents through an internal dashboard to research companies and assemble or correct their profiles, pulling from public information across the web. New companies can also be suggested by users, who can flag errors on existing entries. That process is still rough: one commenter pointed out that a startup's funding figures were badly out of date, citing a far lower valuation than the company's actual, much larger recent round. The creator acknowledged the error, attributing it to agents not searching with recency-specific terms, and said a fact-checking agent and updated prompts were being rolled out to catch similar mistakes.

What Happens Next Depends on the API and Community Input

With only a small number of companies listed so far, much of the near-term roadmap depends on outside input. Commenters suggested scraping investor portfolios, including Y Combinator's public company list, to expand coverage quickly; the creator said batches of roughly 800 YC companies would be added daily over the following week. Other suggestions included open-sourcing the codebase and publishing regular data dumps so the directory's information stays accessible even if the project stalls. None of those changes were live at launch, leaving the platform's longer-term reliability and scope still unproven.