Skip to content

Case Study

Turning a Static Index into a Maintainable AI-Assisted Publishing Workflow

A publishing system that keeps notes, sources, and draft output aligned without forcing a CMS.

Client
Independent product build
Role
Information architecture, frontend engineering, workflow design
Duration
3 weeks
Published
2026-05-01
Next.js
TypeScript
MDX
OpenAI API
Vercel

Evidence layer · Context

Where the work started

The content surface started as a flat list of notes and references that was hard to search, update, and reuse.

Evidence layer · Problem

What needed to change

Publishing required too much manual assembly, so writing and reference material drifted apart.

Evidence layer · Constraints

What shaped the solution

  • Keep the setup local-first
  • Avoid a CMS before the editorial model stabilizes
  • Make AI output reviewable before publishing

Process layer · Process

How I moved through it

  1. Mapped content primitives and writing states.
  2. Defined a narrow schema for topics, sources, and draft status.
  3. Added AI-assisted drafting with review points.
  4. Kept output compatible with local MDX files.

Process layer · Solution

What shipped

Built a structured index and publishing workflow with clean content boundaries, making edits predictable and reviewable.

Outcome layer · Result / Impact

What changed

Writing and reference material could move through one simple surface instead of being split across ad-hoc files.

Less time reshaping content; more time refining the actual ideas.

Outcome layer · Reflection

What I learned

  • Small schemas stay editable longer than broad CMS models.
  • Workflow clarity matters more than fancy automation.

Related Project

Kizuna Index

A living index that turns scattered references into a maintainable publishing surface.

View project

Services Involved

Content Systems & Local Editors
Personal / Brand Websites
Back to case studiesDiscuss similar work