This page has been translated into English by machine.

The 4-Quadrant Process: An Operating System for Coherence

First published · Updated

Written for anyone overwhelmed by AI in marketing. Not here: why it works underneath. The Trust Algorithm

The 4-Quadrant process is easy to admire on a quiet day. Four steps of a loop, Q1 human strategy, Q2 AI execution, Q3 human validation and Q4 deployment, each with a non-negotiable rule about who leads, and on a day when nothing moves any arrangement of them looks fine. Whether an operation actually runs the loop shows on the day something changes.

Q1: a decision, written down

Human strategy is the mode where a person decides what is true and what the business believes, and writes it down somewhere a machine and a checker can both read it. That place is the System Seed. A voice note is raw material, and the decision is the set of lines that go into the seed, each one traced back to something someone said or typed, with the date it was said.

The difference matters because a machine cannot tell a mood from a decision. Hand it the voice note and it will write a confident page on the subject, filling every gap with whatever the rest of the internet says about it. Hand it the seed and it gets the specific claim.

Q2: every page that says the old thing

AI execution starts with a list. Beside the seed sits a record of which pages carry each fact, so a change to one fact produces, on its own, the list of pages it has made stale. Nobody has to remember where the old description lived.

Then the machine rewrites each page from the seed and the page as it stands, so that nothing true is lost and nothing stale is carried forward. When a page needs a claim the seed cannot support, the work stops there and the claim comes back to a person as a question rather than going out as a sentence. The machine executes. It does not get to decide what the business believes in the gaps.

Q3: checked against the source, not the sentence

Human validation is the step most AI operations skip to save time, and the scaling of errors is what they buy with the time they save. The check starts at the seed: a person reads every line of a change before anything is written from it, and reads the output against the seed before it ships.

What makes the check hold is that a fix never goes into a page. If a sentence is wrong, the error is in the seed or in the rules, and correcting it there corrects every page that reads from either.

Q4: deployed together, or visibly not yet

Deployment is the moment the checked pages go live, and the pages one change touches go live together, so a reader moving between them does not meet the old description on one page and the new one on the next.

Not every stale page is rewritten in the same sitting, and that is where the record earns its place. Until a page's update is live, it stays in the queue, marked stale and waiting, rather than sitting quietly wrong with nobody aware of it. Without that list a change opens a period of inconsistency that nobody can measure, because nobody knows how many places still say the old thing.

The loop closes in the seed

The loop is drawn as a cycle rather than a line because deployment teaches something, and what it teaches has to get back to Q1. When a published page and its own source disagree, the decision goes into the seed rather than onto the page, and every page that reads from it joins the queue, to go through Q2, Q3 and Q4 again from the corrected source.

A fix made only on the page would look the same to a reader that afternoon. It would also leave the seed saying the opposite, ready to put the old fact back into the next page written from it.

What the rhythm runs on

Q1 to Q4 is the operating rhythm, and on its own it is not the whole operating system. The rhythm runs on two layers, what you know, which is the seed, and what is true about the business while it runs, which is the core, the customers, products and jobs and the history of every change to them. A change reaches every page that needs it only when the knowledge layer already exists and every page already reads from it.

A business whose prices live in a spreadsheet, a CRM and a pile of hand-written copy can run the same loop, but every pass starts with a hunt for where the old fact is written, and the hunt is where coherence leaks away. Cascading Updates is the workflow for running that hunt well until the single source exists, and the framework puts the loop inside the four quadrants.

The next time a fact in your business changes, write it in one place before you touch anything else, then list every place it has to reach. If you cannot produce that list, the process has a Q1 and a Q4 and is missing the part in between that makes them one system.

Some links on this page pay me a commission if you buy. How these links work, and what I earn, is at timmybrown.co.nz/links.