Evidence trail
The site is built from two local documents: the raw transcript as primary source and the CKF white-paper summary as secondary source. This page preserves the core evidence discipline.
2,284-line raw chat transcript containing founder prompts, corrections, approvals, assistant synthesis, and the CKF prompt.
393-line CKF summary with source index, glossary, evidence ledger, risks, KPI ladder, operating cadence, and QA pass.
Corpus handling
The raw transcript carries founder intent, corrections, and approvals. The CKF summary is used to structure the system, but claims are kept lighter where the source says they are inferred, asserted, or unknown.
2,284-line raw chat transcript containing founder prompts, corrections, approvals, assistant synthesis, and the CKF prompt.
393-line CKF summary with source index, glossary, evidence ledger, risks, KPI ladder, operating cadence, and QA pass.
Evidence sample
A full evidence ledger exists in the CKF source. This site uses the same standard so the OS does not quietly turn guesses into doctrine.
Repeated founder clarification in the raw transcript and CKF.
Approved across the raw transcript and summary source index.
Explicitly accepted by the founder in the raw transcript.
Consistent across the phase roll-up, evidence ledger, and founder response.
Favoured in the transcript, but final public naming is not completely locked.
Reasoned from client examples and behavioural fit, not quantified proof.
A founder belief in the corpus; not yet proven by quantified outcomes.
The corpus has measurement intent but no final metric dictionary.
Audience logic
This is one of the most useful communication rules in the corpus: do not show every audience the whole machine.
Show: Problems solved, service pillars, outcomes, industries, proof, and next step.
Keep internal: Atomic Flywheel, Dragonfly, HyperCTRL.ai, and deep agentic workflow logic.
Show: Their rollout plan, scope, dashboard tier, inputs required, and expansion logic.
Keep internal: Internal architecture detail unless it improves trust or decision-making.
Show: Product pillars, workflow modules, SOPs, naming conventions, QA standards, and handoffs.
Keep internal: Only sensitive financial, IP, or client-confidential strategy.
Show: Full task-relevant context, source map, definitions, constraints, and acceptance criteria.
Keep internal: Anything irrelevant to the specific task.
Missing inputs
These are the inputs most likely to change implementation decisions or block safe automation.
Formal North Star metric and denominator rules.
MQL, SQL, opportunity, qualified lead, and attribution definitions.
Call intelligence consent, privacy, and storage governance.
Dashboard schemas and data ownership rules.
Pricing and packaging for modules, retainers, and advisory.
Threshold for industry page versus vertical arm versus separate brand.
Final public name and scope for personal-brand/thought-leadership work.
Quantified, approved case-study outcomes.