What this case covers
Over about two months, using a seven-day holiday, I broke my personal content creation process into material collection, topic selection, material packages, first drafts, manual revision, review, and Skills, managing it together with Codex and Obsidian. When things go smoothly, daily output of satisfactory drafts increases from 2–3 in the past to 8–10.
Questions it answers
- Why should the creation process without AI be broken down first?
- How do Codex, Obsidian, rule files, and Skills divide the work?
- How should material completeness and viral-hit patterns be handled separately?
- Why should the author first verbally explain viewpoints and experiences before producing a first draft?
- When is it appropriate to solidify the process into Skills and scheduled tasks?
Our observations
- The productivity increase is built on the author's existing products, content, and revision judgment, not an automatic result after installing tools.
- Keeping AI first drafts and my own revised drafts saved separately is what allows real preferences to become input for the next creation.
- Automation should happen after the process is stable; stopping when the previous step fails is more reliable than continuing to generate with empty data.
From the author
The business is changing, but I have always been doing the same thing: continuously turning my own experience, products, and judgment into content others are willing to read.