Stop 7 of 9 · Weekly
Cowork runs the batch
Six approved masters become twenty four derivatives, delegated with a brief that says what to flag instead of decide, reviewed against a manifest, with the judgment kept.
The line the batch does not cross
Stop 2 produced thirty files from six masters. The six are Farah's afternoon. The twenty four are four hours of resizing, reflowing, renaming and exporting that nobody has ever enjoyed, and that is the shape of work Cowork was made for.
The line is simple enough to say in one sentence: multiplication is delegated, judgment is not. A derivative of an approved master is multiplication. A new Arabic line, a rights call on a new image, a crop that cuts a logo or a face, and the final look of the set are judgment.
The delegation brief
A batch brief has six parts and fails when any one is missing.
| Part | What it says |
|---|---|
| Inputs | The master files, the matrix rows, the export list, the folder |
| Rules per format | Safe areas, minimum type size, logo clear space, colour space |
| Flag instead of decide | The list of situations that stop a file rather than solve it |
| Naming and manifest | The file name pattern and the manifest columns to produce |
| Stop conditions | What ends the run entirely rather than flagging one row |
| Out of scope | New copy, new imagery, rights calls, anything not on the matrix |
Here are the six approved masters, the asset matrix rows 7 to 30, and the export list: [paste the three]. Write the batch brief for a Cowork task that produces the derivatives. Cover inputs, the rules per format, what to flag instead of deciding, the file name pattern, the manifest columns and the stop conditions. Say explicitly that no new copy is written, no new imagery is generated, no Arabic line is changed and nothing is posted anywhere. Then list the five ways this brief could be misread and tighten each one.
Those last five lines are the part worth the time. A brief that survives being read badly is a brief you can hand over twice.
The review pass
[attach the manifest and the flag list] Compare the manifest with the asset matrix. Tell me: which rows are missing, which files exist that no row asked for, which file names break the pattern, and which flags block Sunday against which are cosmetic. Group the flags by cause, and for each cause say whether it is a defect in the master, in the brief or in the format rules.
From the flag list, give me two lists. One, what I fix in the master and re-derive. Two, what I fix in the brief and re-run. Order both by how many files each fix clears, and name the single change that clears the most.