Meet the Builder — An AI Gemba Walk
CRA-001 — The Model Is Not the System
The Evidence Library for the current claim boundaries
You do not need to read everything. Follow the part of the map that answers the question you actually have.
What is supported today?
The public material distinguishes between several evidence levels instead of treating everything as simply “verified.”
Architectural claims: how the runtime is designed and how responsibilities are separated.
Implementation evidence: code, configuration, and public technical documentation showing that components exist.
Executed evidence: controlled runs and traces showing that selected paths operated under specific conditions.
Production observations: behavior observed in the current runtime, where that evidence has been preserved.
Independent replication or third-party audit: future work unless a document explicitly says otherwise.
The Nexus audits are self-administered engineering evidence. They are useful starting instruments, not independent certification.
WHY
The story and mental models behind the work. It explains how manufacturing, logistics, SQL, quality systems, and relational learning shaped the way I approached AI.
Approx. reading time: 5 minutes on the site; longer for the book
ARCHITECTURE
Continuity Runtime Architecture explains the responsibilities surrounding the model: identity, memory, context assembly, governance, tools, persistence, maintenance, and continuity.
Approx. reading time: 15–35 minutes per paper
INSPECTION
The Continuity Runtime Audit Methodology separates what an architecture proposes from what an implementation contains, what actually executes, and what has or has not been independently validated.
Approx. reading time: 20–45 minutes
REFERENCE IMPLEMENTATION
Nexus is where the ideas were built, tested, broken, repaired, and revised. It is one implementation, not a claim that every continuity runtime must look the same.
Approx. reading time: 25–35 minutes