The build log
One human editor. An AI newsroom. Public numbers.
This publication is itself an experiment: how much of a real trade publication can an AI editorial system run, with one human editor spending about thirty minutes a day? This page is the lab notebook. The numbers below are real and update monthly.
The machine
- Scan. Every morning, an automated pipeline reads primary sources (company pressrooms, earnings transcripts, trade press, operator communities) and ranks the stories where something actually changed for marketing teams.
- Draft. An AI editorial system builds a fact sheet from the sources, then writes the article against this publication's standard: verbatim linked quotes, a visible source list, a clear argument.
- Gate. A human editor reads every draft and approves, edits, or kills it. Nothing publishes without sign-off.
- Distribute. The same pipeline drafts the LinkedIn post and the newsletter blurb. The editor presses the buttons.
The numbers
| Month | Articles | Editor hrs/week | Subscribers | Notes |
|---|---|---|---|---|
| July 2026 | Launch week | ~3 | Counting from zero | Site, newsroom pipeline, and first stories shipped. Numbers start accruing now. |
Why we publish this
Because it is the most honest sales page we could build. JAC Growth Marketing builds content engines like this one for B2B companies. Rather than tell you it works, we run one in public and post the numbers. If the numbers stop being good, you'll see that too.
Watch the experiment run.
The briefing on what AI is doing to marketing budgets, teams, and careers. News and analysis, twice a week, five minutes.