Training AI to Predict the Future

Our novel self-play framework* learns directly from real-world feedback.

No curated data.
No humans in the loop.
No bottlenecks.
Any domain.

AlphaGo for the real world.