Autonomously create
knowledge.
We're building autonomous systems of models that form hypotheses, design experiments, and arrive at discoveries no one has made before. Our systems have already invented new ML methods and algorithms that significantly beat the state-of-the-art on different ML problems at a fraction of the cost.
Not just agents,
but scientists.
Today's AI agents are capable at orchestrating and executing complex tasks. But the hardest part of building new software isn't writing the code. It's knowing what to build and why.
At rekursiv.ai, we're working on a more capable form of intelligence which doesn't just follow orders, but acts as a scientist by hypothesizing, designing experiments, running them, interpreting results, and discovering new knowledge.
Our AI Scientists have already surprised us many times. They've discovered novel algorithms to real problems through thousands of experiments and at a fraction of the cost of human research.
And we're working to scale this up to millions of scientists, producing billions of discoveries.
Read more about the first AI Scientist system and the algorithm it discovered →
Built by researchers,
for researchers.
Our founding team brings almost 30 years of combined research experience at Google, DeepMind, and Luma AI. Between us, we built TensorFlow Probability, led the models powering Dream Machine, and developed VideoPoet (ICML 2024 Best Paper). Now we're applying that experience to a harder problem: AI that discovers on its own.
Join Us →
