AI agents trained on Fortnite: General Intuition raises $320 million

General Intuition figured out something simple: an agent playing Fortnite faces the same chaos as the real world. It must decide in real time with incomplete information while everything changes. Gameplay data is pure: concrete actions, immediate consequences, instant feedback. Not theory on paper, actual battle.

The company is using millions of hours of skilled player footage to teach models something like human intuition. It's not rule memorization, it's pattern recognition in chaotic situations. General Intuition raised $320 million to scale this approach and bring it into physical robots.

That's where the magic gets concrete. An agent trained on Fortnite learns to plan several steps ahead, manage uncertainty, adapt when things go wrong. These transferable skills are exactly what a robot needs to grab objects in a messy warehouse or navigate space it's never seen before.

Traditional AI learns from labeled lab data, sterile and predictable. Gameplay is wild: thousands of variables, unrepeatable situations, suboptimal choices that still work. This teaches algorithms the flexibility that perfect datasets never will. The leap is huge: from controlled benchmarks to real-world chaos.

Because robots become actually useful when they understand the world like we do. General Intuition isn't building AI that plays Fortnite better. It's building practical intelligence that'll cut training time for real physical tasks by years.

Investors saw the potential: this approach could accelerate robotics compared to traditional methods. It's not a game. It's the next step toward AI that doesn't need artificial environments to learn.