What's here
Posts on getting real work done with LLM agents. Not tutorials. Not benchmarks. The day-to-day texture of running workflows where the model is a collaborator and the human's job is to notice when the work goes sideways.
editorialcontrol.org
A publication about building with AI agents — workflow patterns, editorial cadence, and the automation that actually ships, written by someone who would rather be using the work than describing it.
Editorial spread
Why bother
Because the interesting part isn't the model — it's what it does to your process. Calendars, review cycles, publishing cadences: all of it shifts when the person at the desk can delegate, then correct. That shift deserves a press of its own.
How we publish
Slowly. Every piece here is drafted by an agent and finished by a human, using the same editorial calendar you can read about in the posts. Nothing ships that wouldn't survive a read-aloud. The receipts stay on the page.
On the desk
The running list of sections this publication is working on. Each one is a place we expect to return to, not a one-shot post.
- § 01
Workflow patterns
What it looks like when agents do the grunt work and humans review. Templates, branches, review loops.
- § 02
Editorial cadence
Shipping writing on a schedule. Calendars, drafts, review, publication — run by agents, finished by a human.
- § 03
Automation that actually ships
The difference between a cute demo and something you use every week without noticing.
- § 04
Instrumenting the loop
Analytics, evals, audit trails. What feedback an agent needs to get sharper run over run.
- § 05
Field notes from the desk
Specific sessions where the workflow worked, broke, or surprised us. No generalizations without a receipt.
The dispatch
When the first issue ships, it goes out to a short list. Want it?
Write to the editorNo form, no funnel. An email. Reply asks to be added; I reply when the first piece is close.