Thinking
Knowledge becomes an asset only when it is made public.
This is where patterns of judgment developed through client work are brought outside. It is not a display of ideas. Only records of decisions actually run, and methodologies drawn from them, belong here. There are no signs for domains that have not been practiced.
Why publish?
Not because we want to communicate.
It starts with one question: “Is my tacit knowledge worth putting outside?” If it is, it should become an asset rather than be allowed to decay. That is close to an obligation.
The place is also decided. Not the saturated field of “AI development techniques,” but territory that no one has yet named: AI governance and organizational authority design. Rather than become the nth voice in a red ocean, plant a flag in open space.
There is only one moat to protect: Evidenced × Opinionated. Evidence from work actually run, and the posture of stating a view plainly. Authority cannot dig this moat. That is why failures are written too. Persuasiveness comes not from success stories but from failure, and that is how this place is held as a practitioner rather than a commentator.
And do not exaggerate. Apply the discipline of decision quality most of all to what is published. Write weaknesses without inflating them. Exaggeration works in the short term and erodes the moat in the long term.
Relationship to Qorum Method
At the core of this page is Qorum Method.
It is not a methodology for deciding what to entrust to AI. The opposite: draw in advance, by risk level, the range AI must not be allowed to decide. At its center is organizational authority design itself: who or what may decide what, at what level of risk. The name comes from that. Its quality is a disruption of quorum: not deliberation, but governance built in. The idea is embedded in the spelling.
The full methodology is published as OSS (GitHub: shiyamaz/qorum-method). Tacit knowledge developed through client work is made into an asset through one continuous pipeline: idea → naming → OSS publication → multiple audits → working tools → diagrams → communication. That is how work is made here.
But claims do not run ahead of evidence. They do not float in excessive abstraction. Development examples remain as anchors, and as evidence accumulates, the reach of the claims is raised in stages: from governance to PM, PMO, audit, management meetings, and boards of directors.
Four channels, four roles
Each channel has a distinct role. Choose for reach and trust, then tie the paths into one.
| Channel | Role | In one phrase |
|---|---|---|
| A place to communicate who this is to management; profile design that establishes “the person for decision-making governance in the AI era” in five seconds | An engine for acquiring work, and a living asset | |
| X | An entry point for discovery. Briefly isolate one issue from published judgments and practice, then connect to the site, LinkedIn, and note | A short point of contact |
| GitHub | Implementation of the methodology: the kernel. Publish the full workings and make verifiability itself the evidence | The core that does not grow bloated |
| note / serial writing | A layer for developing ideas. Expand through domain-specific playbooks without bloating the core | The layer that creates breadth |
From the body to the destination, and from the destination to deeper assets. The end point of the journey is deliberately one.
What will be added here
What appears here is not opinion, but records of judgment.
More records of judgment will be added here. We write only as much as we have run. What we have not run, we do not write.