Make dependencies explicit.
Represent the relationships behind a proposed action so the governing logic has something concrete to inspect.
INDEPENDENT AI GOVERNANCE RESEARCH
Intelligence needs
a geometry
of accountability.
DAXDA explores how explicit constraints, geometric representations, and traceable evidence can govern what an AI system is allowed to do.
FIELD STUDY 001GEOMETRIC GOVERNANCE / DAXDA
FOUNDED BY NICOLE BESSOPEN RESEARCH. INSPECTABLE EVIDENCE.SCROLL TO EXPLORE ↓
01 / THE IDEA
A decision should carry
more than confidence.
It should carry a reason it was permitted.
DAXDA is a research and engineering effort focused on reasoning governance: a layer that evaluates proposed actions against explicit rules and records the basis for its decisions. Its work connects neural-symbolic dependency structures, geometric representations, and audit evidence.
The central question is practical: can a system make its authority to act inspectable and reproducible?
02 / THE PROBLEM & THE APPROACH
From an answer
to an accountable action.
Generating an output and authorizing its use are separate engineering responsibilities.
Evaluate permission to act.
Apply explicit governance conditions at a decision boundary, with a research focus on deterministic, inspectable checks.
Leave evidence behind.
Connect decisions to evaluation records so engineers can investigate failures and attempt independent reproduction.
WHAT HAS BEEN BUILT
The Next Gen READMEs describe a governance engine, Guard SDK, distributed validation components, and deployment tooling. The v11.4 rebuild kit provides a separate entry point for baseline reproduction. These are research artifacts; this page does not assert independent validation, universal AI containment, or production safety.
03 / EXPLORE THE SOURCE
Three entry points.
One research direction.
Read the code. Follow a branch.
Challenge the assumptions.
Branch snapshot: September 9, 2026. “All branches” opens the current GitHub listing. Next Gen repositories are distinct projects; shared README text does not establish identical code.
04 / WHO THIS IS FOR
Build it. Test it.
Help define what comes next.
Reproduce before you rely.
Explore governance logic, inspect evidence paths, test adversarial cases, and identify where the implementation diverges from its specification.
Find your starting point ↗Explore an explicit control layer.
Discuss research collaborations around action authorization, auditability, and evaluation. Suitability for any deployment requires workload-specific testing.
Discuss a use case ↗Assess the evidence and the opportunity.
The thesis: increasingly capable AI needs inspectable controls. Engage on independent validation, integration cost, customer requirements, and a credible path from research to product.
Meet the founder ↗05 / ABOUT DAXDA
Independent by origin.
Collaborative by ambition.
Founded and developed by Nicole Bess.
Nicole is an independent researcher and the sole founder of DAXDA. Her work explores how reasoning systems can be governed through explicit constraints, structured evaluation, and evidence that others can inspect.
DAXDA brings that inquiry into code. The invitation to engineers, research collaborators, and investors is straightforward: examine the artifacts, ask difficult questions, and help test what holds up.
Nicole on GitHub ↗06 / START A CONVERSATION
The next connection
could change the field.
Engineering collaboration, research questions, or investment discussions.
osmesirius@gmail.com ↗Nicole Bess · Founder, DAXDA
Submit a question
Prepare a message to Nicole. Your email app opens for you to review and send it.