Work in progress · duh.bet

Software that remembers what matters.

NeuroSyn Labs is exploring a different kind of software architecture: stateful, adaptive systems built around explicit memory, provenance, topology, and lawful change. The work is early. The claims are being tested. The interesting part is that the pieces are starting to move.

DUHBLEStateful execution + memory research
.duhPortable model format under development
Local-firstDesigned to keep state close to the system
Proof-drivenClaims follow tests, not the other way around
The premise

What if software did not need to fake continuity?

Most software treats memory as something bolted on later: a database, cache, prompt history, or vector store. NeuroSyn is investigating systems where state, conduction, and change are part of the model itself.

State belongs somewhere real

Persistent behavior should come from mounted state, not from a second hidden source of truth.

Change should have a cause

Updates should be traceable to an allowed source, path, and event instead of ambient authority.

Tests should be able to kill the idea

We are actively looking for failure modes, contradictions, and architecture that cheats.

Research / prototype stage

This is not a finished product.

NeuroSyn and DUHBLE are under active development. Interfaces, terminology, architecture, performance characteristics, and implementation details may change as testing continues.

Where it could go

Possible implementations

The architecture is being explored as a substrate, not as one single app.

Adaptive game characters

Characters whose relationship state can persist, decay, reinforce, and respond to prior interaction.

Local assistants

Device-resident systems that can preserve meaningful state without giant remote context windows.

Interactive simulations

Controlled systems for testing emergence, provenance, topology, timing, and state transitions.