The content machine. Yours to take.
The actual pipeline behind the talk: one AI writer that sources a topic, checks every claim against a primary source, writes it in a real voice, critiques itself, and learns from your edits. A human approves before anything ships. Below is the sanitized foundation, free, no email.
Build the hard part once. Reuse it everywhere.
Verify, voice, judge, and learn are built once. Everything else is a thin adapter. The kit gives you each step as a prompt you can run today, plus the poster variant (news → verify → draft → review → publish) and the quality rules that keep it from sounding like a robot.
Grab the whole kit.
Download the full kit (.zip) →
Or grab the pieces individually:
- README — the whole pattern + how to run it README.md
- 01 · Oracle — pick what's worth saying 01-oracle.md
- 02 · Interview — pull the real story 02-interview.md
- 03 · Draft — write it, transcript-bound 03-draft.md
- 04 · Validate — kill the AI slop 04-validate.md
- 05 · Council — four judges, anchored rubric 05-council.md
- 06 · Lessons — learn from your edits 06-lessons.md
- Poster — news → verify → draft → publish linkedin-poster.md
- Quality rules — tightening + attribution quality-rules.md
Built with this, shipped this week.
The pipeline drove +291% LinkedIn impressions in a week (per LinkedIn analytics, Jul 17–23) and got the first-mover take out on a same-day model release. Same-day, verified, in-voice. See the blog and what a fractional AI officer sets up for where this goes at scale.
Straight answers.
Is this really free, no email?
What do I need to run it?
What's intentionally left out?
quiet growth.
Built by Josh Cruz. If you build something with it, I'd genuinely love to see it.