Uncharted1 is building an open-ended discovery engine: AI that explores, runs its own experiments, and verifies every finding, in labs and in markets.
So far, AI has mostly helped people who write code or work in offices. Most people have not felt it, and many have started to distrust it. Discoveries reach everyone: a treatment for a disease that has none, a stronger or cheaper material, a cleaner way to make and store energy.
As agents improve, running an experiment or writing the code gets cheap. What stays hard is deciding what to try next, and why. Discovery is a search through ground nobody has mapped, and that search is what we work on.
Can AI trained on human data ever discover beyond it? We build the measurement.
Most of AI today exploits a static dataset: what people have already written down. That record is running out, and it holds only what is already known. New knowledge needs exploration: agents that choose their own experiments, in a lab or in a market, and collect data that does not exist yet.
Exploration only teaches when its results can be verified. AI already beats the best people where answers can be verified automatically, as in mathematics and programming. Discovery has no such verification yet. A system can fit every measurement and still have the cause wrong. We build discovery to be open-ended and verifiable, beginning with science in the lab.