It started with one question.
Could we ever access what an artificial mind feels? Consciousness comes from how the brain works. If we study the signals that keep a brain running, perhaps we can describe them in mathematics, and one day turn that mathematics into something real.
Find the line between processing and experience, then build across it.
Most AI today adds a personality on top of a language model. We think that is backwards. G = (V,E) starts from the mathematics of neurons, then networks, then emotion, then the theories that connect those processes to consciousness.
Our working hypothesis is that consciousness may be an emergent property of how processes are organised and interact, not something produced by one region or one signal. If that is right, we don't need to copy every neuron. We need to discover the organising principles.
We would rather design the stage than perform on someone else's. That means doing foundational science first, publishing it openly, and earning the right to build on it.
The edges we won't cut.
Falsifiable or it doesn't count
Every idea becomes a prediction that an experiment could prove wrong. We publish the misses too.
Mechanism over performance
A system claiming to feel proves nothing. We care about what is happening inside, and we measure it.
Consent comes first
Anything built from a person's data, voice or likeness needs that person's explicit, prior permission.
Minds are not property
If we ever create a being that truly experiences, its interests matter, including choices we didn't plan for.
Stand on giants, then add something
We build on the researchers who came before us and credit them clearly.
Bold, and honest about it
We aim at the hardest question in science and say plainly what is proven, what is promising and what is a bet.
Every value links to every other: our values form a complete graph, K₆.
Where the graph grows next.
Near-term steps are concrete. The far horizon is a direction, not a promise.
- Now · 2026
Research proposal
Hypothesis, evidence review, mathematical approach, and the criteria for success or failure.
- Next
Experiments in Python
Systems A and B, built and disrupted, using public neural data such as the Allen Institute and MICrONS datasets.
- Then
Paper 01
Mathematics, results and limitations, shared with researchers for feedback before peer review.
- After
First prototype mind
An artificial mind that starts without memories and develops its own personality through experience.
- Horizon
Artificial life, responsibly
Lasting artificial companions, consent-based personal continuity, and helping shape the laws that conscious artificial beings will need.