This note records two things that happened in sequence. First, I completed the 970LAB AI content workflow on a Mac mini: topic selection, writing, images, human review, AI narration, finished video and publishing. Later, I completed automated publishing for my Guoma970 account on a MacBook Pro. Each project addressed its own problem, on a different device and account.
The Mac mini work covered the full 970LAB content production workflow. AI helped generate and organise material, while human review remained at key points. The later MacBook Pro work concerned automated publishing for Guoma970. Each had its own output.
These experiences are the starting point for writing reusable instructions. The public ai-content-ops package already defines how to hand over sources, frozen versions and asset mappings, with separate review and authorisation gates. I can compare my local practices against those conventions. Sanitised examples, configuration instructions and a clean-environment run are still missing. The work happened; packaging it for others is only beginning.
Two experiences, with separate results
Completing the 970LAB workflow on a Mac mini and later automating Guoma970 publishing on a MacBook Pro were two separate pieces of work. They were not two runs designed to test portability.
The first produced a complete content workflow; the second produced automated publishing for another account. Describing them as the same workflow running twice would lose that context.
Public conventions and local experience
The public ai-content-ops package describes verifiable handoffs: sources and approved versions, asset mappings, review and authorisation status. It does not log in to accounts, upload or publish on someone’s behalf.
Those conventions provide a reference for reviewing the local work. This article does not claim that every convention was verified in both projects.
I completed the local work using my own devices and accounts. The public package itself does not operate accounts.
What still needs to happen
First, prepare sanitised input and output examples that can be shared publicly.
Second, explain which parts are general methods and which depend on local device or account configuration.
Third, run the process in a clean environment without private local folders or existing account sessions. Until then, reusability and out-of-the-box operation remain unproven.
AI-assisted writing and illustrations. Narration uses an authorised synthetic version of my own voice. Illustrations explain the process; they are not software screenshots.