Eusophic
Try NY ↗

We build minds that have to learn

Eusophic is a research and technology company. We design cognitive systems that acquire what they know through experience, and the worlds they acquire it in.

Our purpose

Learning from experience

We build cognitive systems that start with nothing and acquire language, memory and goals from what happens to them.

Evidence as a method

We fix our thresholds before an experiment runs, tie each result to reproducible artefacts, and report what is still unverified.

Products you can inspect

What we learn, we release as software that runs on your own machine and can be read at any time.

A world to learn in

Learning needs somewhere to happen. We design consistent simulated worlds that set the rules and never choose for the mind inside.

What we have released

NY, short for Natural Yield, is our first product and our research platform: a native cognitive system living in a persistent simulated Manhattan. It is in an invite-only alpha.

NY · Origin instance
Places known 01
Routes remembered 00
Illustrative

What we value and how we act

Building a mind slowly, from experience, means choosing again and again not to take the shortcut. These principles are how we make those choices.

Earned, not installed.

We would rather a system know a little that it learned than a lot that it was given. An Origin instance starts with nothing, and every later piece of knowledge has a traceable cause.

Evidence before claims.

A capability is described only as far as it was tested. We write down what is verified, what is bounded and what is still open, and keep the three apart.

Small enough to inspect.

We prefer systems a person can hold in their head. NY runs on an ordinary CPU in tens of megabytes, and its goals, learning and memory are readable at any time.

Private by construction.

What an instance lived through belongs to that instance and its owner. It stays on the owner’s machine, and is excluded by design from anything that can be exported.

No borrowed mind.

We do not place a pretrained model behind the curtain. If our systems appear to understand something, that understanding was built here, from experience.

Say what it is not.

We state plainly what our systems are not: not a language model, not a chatbot, not a claim about consciousness, and not connected to real money.

Why Eusophic

The name joins two Greek roots: eu, good, and sophia, wisdom.

It is a reminder of what we are aiming at. Not systems that simply know more, but systems whose knowing is sound: acquired honestly, held with the right amount of confidence, and open to correction.

We do two things. We study how a machine can learn from its own experience, and we release what we find as products that people can run, watch and question.

Contact

[email protected]

Questions, feedback and collaboration are all welcome. For access to NY, join the waitlist at ny.mba.