Abstract This paper documents the journey by which Cognitive Memoisation (CM) was articulated, strengthened, and governed through extended interaction between a human practitioner and a Large Language Model (LLM). Rather than presenting CM as an invention or a finished framework, the paper describes a process of reflexive bootstrapping in which an already-existing human practice was progressively rendered explicit, constrained, and stabilised. The LLM functioned not as an epistemic authority, but as a low-impedance cognitive surface capable of carrying ambiguity, reflecting structure, and probing boundaries. The paper shows how terminology, invariants, and governance emerged through use, correction, and externalisation, and how epistemic authority remained explicitly human throughout. The result is a case study in disciplined human–machine collaboration that enables progress without memory while avoiding semantic drift and authority transfer.