ORIAS
ORIAS is an independent research and engineering lab. We look for new sources of capability in mathematics, intelligence, computation and physics, and build toward machines that can do what today's cannot.
We want to know how much of what machines can't do is a law of nature, and how much is only a gap in what we've understood. Then we want to close the gaps that turn out to be ours.
Disciplines aren't departments here. They are places where a powerful principle might be hiding.
Mathematics Plate II
Not only using the mathematics we have, but questioning the language problems are written in. Some apparent limits belong to the description, not to the world.
Melencolia I, Albrecht Dürer, 1514. A mind, a polyhedron, a magic square, and a problem it hasn't solved yet.
Intelligence Plate III
Not just larger models or longer reasoning. Learning, memory, discovery and understanding, and forms of cognition that may look nothing like a human mind.
Radiolarians, Ernst Haeckel, 1904. Nature found geometries nobody had thought to draw.
Computation Plate IV
The boundary between abstract mathematics and physical machinery: shortcuts, unexpected capabilities of existing hardware, and new ways of organising computation itself.
Joseph Marie Jacquard, whose loom of 1804 read punched cards: a machine that followed a program.
Physics Plate V
Measurement, dynamics, thermodynamics and quantum systems: where the hard limits really are, and where physical processes might compute in ways we haven't used.
Johannes Kepler's model of the planets, 1596. Beautiful, precise, and wrong. He found the right laws later.
Not a line of finished products. Different attempts at hard questions, each with its real status.
Experimental software intelligence: software that could be inspected through alternative runs, changes and accumulated evidence. It taught us that holding information is not the same as understanding it.
Machine-native scientific reasoning: executable theories of small worlds, compared and told apart by experiment. Laplace's demon, minus the omniscience.
A working research harness for structured investigations, experiments and verification. A tool for doing research, not a claim of autonomous discovery.
Mathematics of what an observer can and cannot infer about the process behind what it sees, including systems that look reversible from outside while dissipating energy inside.
Our open research initiative: the search for new sources of capability, which may turn out to be a machine, a principle, or something we don't yet have words for.
An ORIAS research initiative
Somewhere past the edge of what today's machines can do, there may be principles nobody has found: a new kind of operation, a new kind of intelligence, a new way for physics to compute. Worldfold is the search for them. We don't assume we know what we'll find.
Talk to us about Worldfold
We are trying to discover what deserves commitment, not to defend the first idea that sounded good.
Ideas become breakthroughs through real mathematics, real engineering and real experiments, and we say which stage each one is at.
A new measurement, a new mathematical operation or a new physical mechanism can matter more than a smarter model.
Software, mathematics, existing computers and rigorous experiments are where every frontier begins.
"We need enough of a vision to begin, but enough freedom to be wrong about almost everything."
From the founding brief, October 2026
Researchers, engineers and partners who care about what machines could become: we'd like to hear from you.
hello@orias.dev