Internal V1.1 test artifact presenting seven reasoning axioms for human-AI work. The axioms are designed as compressed state-change rules that can be pasted as context into ChatGPT or Claude to improve structure, boundary-setting, assumption separation, and decision usefulness. The seven axioms were tested using real outputs from ChatGPT and Claude against five visible improvement criteria. This record is not a scientific validation, AI safety validation, governance validation, benchmark claim, or public product release. It does not claim to make AI smarter.
Honan et al. (2026) studied this question.