It's easy to look at a company's trajectory after the fact and assume the signs were obvious. They rarely are, in the moment. The companies that eventually became unicorns spent meaningful stretches of time looking, to almost everyone outside a small technical community, like nothing in particular — a repository with a modest following, a Hacker News post that got a few hundred upvotes and was forgotten by the next morning, a Product Hunt launch that didn't trend.

What actually distinguishes the pattern from noise

The difference between a startup that fades and one that compounds isn't usually visible in any single moment. It's visible in the trajectory across multiple independent signals over time — sustained contributor growth rather than a single spike, cross-platform corroboration rather than isolated mentions, and novelty scores that stay elevated rather than decaying back to baseline after the initial post.

This is precisely why Signal Intelligence tracks 110 entities continuously across AI & Startups rather than relying on any single viral moment. A single Product Hunt launch tells you almost nothing on its own. The same entity showing sustained GitHub activity, repeated Hacker News mentions, and accelerating contributor diversity over multiple weeks tells you something structurally different is happening.

The three collectors feeding this domain

  • GitHub Trending — repository momentum, commit velocity, contributor diversity
  • Hacker News — front-page appearances and sustained discussion engagement, not just submission
  • Product Hunt — launch activity and follow-up engagement over the following weeks, not just launch day

The OpenAI precedent, again

Twelve entities currently tracked in this domain show signal profiles matching OpenAI's own 2017–2019 pre-recognition pattern — a period when OpenAI itself looked, to most of the public, like a research lab among many. The pattern that eventually became obvious was visible in the signal data years before it became obvious in headlines.