How Artificial Societies raised $5.35M to let founders, creators, and researchers test reality before it happens.
It was a mirror, showing not just how one person might react, but how a large group of them would.
So they leaped toward something radical.
What if you could build an artificial society, a digital replica of the real world that could test how people might respond to decisions, products, and policies before those things even happened?
The results were striking.
In tests, simulations mirrored real-world outcomes. Collective behavior could be mapped, even rehearsed.
That experiment became the seed for a company: Artificial Societies.
In 2024, James He teamed up with Patrick Sharpe to launch it with one audacious goal: build entire societies using AI, and use them to anticipate how humans behave before they act.
The technology runs on what they call synthetic personas. Each AI clone carries traits, preferences, and biases that make it behave like a real human.
One clone is interesting. Ten thousand are powerful. But half a million? That’d be a population.
Feed them a scenario, a new policy, an ad campaign, a breaking headline, and you could watch how reactions ripple through the crowd.
Momentum came fast. In August 2025, the company raised $5.35 million in seed funding (led by Point72 Ventures with support from DeepMind, Strava, and Sequoia Scout alumni) to scale its vision.
By then, 15,000 users had already run more than 100,000 simulations.
Creators testing TikTok hooks and viral memes.
Startups testing their marketing campaigns.
Researchers experimenting with public opinion.
But with the breakthrough came unease.
Because if an AI clone can forecast your next click, your next vote, or your next purchase, does that hand too much power to whoever owns the model?
The founders push back on that idea. Their framing is clarity, not control.
