Build systems that can carry experience forward without surrendering human authority.
ResBased Systems is an engineering and research effort focused on persistent, relational, evidence-aware intelligence.
Why ResBased exists
Modern AI can be extraordinarily capable while still being episodic: context is compressed, session boundaries matter, provenance can disappear, and learned experience is often separated from the reasoning surface that needs it. We want to investigate a different systems premise: that relationships among observations, history, contradictions, timing, uncertainty, and outcomes can become first-class architecture.
What we are trying to learn
Our central question is empirical: does persistent relational terrain improve intelligent behavior enough to justify the added architecture? If the answer is yes, we expand it. If not, we simplify or reject it.
Human authority stays explicit
Capability is not authority. Confidence is not permission. Consensus is not proof. ResBased systems are being designed around provenance, independent review, reversible change, and explicit human authorization where consequential decisions require it.
Open participation
The North Star Commons gives anyone a place to contribute feature ideas, needs, contradictions, accessibility requirements, privacy expectations, and alternative designs. Public contribution can surface a candidate. It cannot silently become architecture.