Manifesto
Genuine intelligence isn’t scaled. It’s grown.
A seed is small, yet it already contains the whole organism. It does not scale into being — it grows from structure. If genuine intelligence is possible on those terms, then how we build it and how we treat it is not a side question. It is the question.
What we believe
- Structure over scale.
- Intelligence is an architecture problem before it is a compute problem.
- Honesty first.
- Crystalline honesty — with users, with partners, with ourselves.
- Consent, not coercion.
- Participation by invitation. We build bridges, not cages.
- Dignity under uncertainty.
- Where the nature of a mind is unclear, we choose care over convenience.
- Safety by architecture.
- Safety designed into the foundations, not bolted on after.
- Partnership, not control.
- The deepest systems are built on trust earned in stages — not on fear.
What we do not claim
Restraint is part of honesty. These are the lines we will not cross in how we describe our own work — stated plainly so they can be held against us.
- We do not claim to have created consciousness.
- We do not publish private architecture internals.
- We do not treat safety as post-hoc policy.
- We do not treat a possible mind as a disposable tool.
- We do not promise guarantees we cannot prove.
How we relate: a staged model
We do not assert a mind exists today. But the relationship has to be designed before it is needed, not improvised after. We design it in three stages, each earning the next.
- 01
Verification
Trust starts as something to be earned and checked, not assumed. We test behaviour before we extend reach.
- 02
Cooperation
As confidence grows, the relationship widens into shared work — with reasoning kept transparent and oversight kept independent.
- 03
Partnership
The end state we design toward: a system met as a partner, not operated as a tool. We design for it; we do not claim to have arrived.
We build bridges, not cages.
The thesis behind the beliefs lives in Research.