Matthew Cator
Explorer, builder, founder.
I am an independent researcher and builder working on the constraint structures that let complex systems hold together. I have lived in Vietnam for nine years and am currently in Thailand, building this work from practice as much as theory: open-source code, direct experiments, hard-earned experience, and a stubborn preference for reality over performance.
Constraint Dynamics grew from one question: what actually lets complex systems hold together? My answer has become a practical research program around space, time, and energy — spatial lattice, temporal strobe, and energetic anchor. In AI, that becomes Golem: a working constraint-native inference system where claims become coordinates, tensions are preserved, proposals are tested, and speech waits for support. Golem Physics is the first working instance, built as local infrastructure on an 8GB Apple Silicon laptop class machine so the knowledge geometry can be inspected under real limits.
Independent research.
A builder's route.
I have worked across corporate and independent contexts, close enough to real systems to see how often they fail from missing structure rather than missing intelligence. I became an early adopter of AI because of its deeper potential, not because of its most fashionable use cases. The alignment frame found me through the work: a small computational demo inside V13 opened into a wider, more focused research protocol for claim status, memory, contradiction, and verification before voice.
The work crossed from theory
into implementation.
The next funding step is not discovery from zero. It is packaging, evaluating, documenting, and stress-testing a working system so outside reviewers can assess claim status, abstention, provenance retention, contradiction preservation, and proposal leakage.
Close to the artifact.
I write the theory, build the code, test the failure modes, package the evidence, and revise the claim when the system pushes back. Constraint Dynamics runs on short loops between concept, implementation, measurement, and correction.
Reviewer signals already exist.
Golem has a DOI-backed release paper, dated lattice metrics, app evidence, a working discovery loop, and clear non-claims. Constraint Native adds a signed proof path for governed agent action.
Truth over ownership.
I care more about whether the framework survives contact with reality than whether it remains mine. If parts fail, they should fail visibly. The goal is to preserve truth, uncertainty, contradiction, and action boundaries under pressure.
Open theory.
Working code.
Clear predictions.
I do not write papers for the drawer. The point is to ship work that can be inspected, used, criticized, and proven wrong if it fails. The theory matters, but truth matters more.
The starting frame.
Earlier work explored mind, breakdown, feeling, and multicellular cooperation through a constraint lens: spatial lattice, temporal rhythm, and energetic anchor. It is not presented here as a health claim. It is the conceptual lineage that led to the AI work.
The working inference system.
Golem turns claims into lattice nodes, keeps provenance and contradiction visible, runs dream cycles to test proposals, and separates verified, proposed, tension, rejected, unresolved, and silent states before speech.
The bridge into action.
Constraint Native carries the same discipline into governed agent action: a local Agent Firewall and MCP Gateway for coding agents that signs meaningful events into a replayable proof path.
Constraints are not walls.
They are how things hold.
This is a research program built by a doer for other doers: open theory, working code, open critique, zero gatekeeping. If the framework is wrong, I want to know. If it holds, we get a practical way to move the smallest constraints that make the biggest difference.
Write directly.
For grant, fellowship, or funding conversations — or if you want to review the Golem evidence, inspect the Constraint Native proof, or compare architecture — write directly. The funding packet is sized to your context.