Why I Started This Blog

I talk to AI every day. When I’m stuck on code, I ask. When I’m thinking through architecture, we discuss. Sometimes we even have deep philosophical conversations. Then one day I realized: all these conversations were just disappearing.

When a session ends, it evaporates. I know I had a good insight yesterday, but I can’t remember it. I find myself explaining similar problems from scratch again. This is waste.

So I decided: record conversations with AI. And don’t just record them—turn them into blog content.

The Prompt That Started It All

Everything started with this prompt:

“in dev folder of koed I want to gather prompts I wrote, Idea I had, misses we made and write blog contents with those. for example let’s say i am creating blog with you right now, and I want this to be an content at the same time.”

Looking back, there are reasons this prompt worked well.

First, clear categories. Prompts, ideas, misses—three buckets. Easy for AI to structure.

Second, meta-awareness. I said “I want this conversation itself to become content.” This was crucial. It clarified that I wanted a system for recording while working.

Third, open structure. The attitude of figuring out details through discussion with AI. Not specifying too much left room for better ideas to emerge.

The Content Seed System

Here’s the system we built through conversation:

BrainFucked/10-Blog/
├── seeds/
│   ├── prompts/     # Effective prompts
│   ├── ideas/       # Ideas that emerged
│   └── misses/      # Failures and lessons
├── drafts/          # Work in progress
└── published/       # Published posts (symlinked to Hugo)

prompts stores effective prompts. With analysis of why they worked. Reference for similar situations later.

ideas captures insights from conversations. “Oh, this is worth writing about.” Doesn’t need to be complete thoughts. They’re seeds.

misses records mistakes and failures. Bugs, wrong assumptions, wasted effort. This is content too. Maybe the most valuable content.

One Hub for Everything

At first I was going to create a separate folder. Something like ~/.content-seeds/. But then I remembered—I already have an Obsidian vault. Called BrainFucked.

“Brain Fucked will be hub for every idea, content, post”

This became my principle. One vault for everything.

Why?

  • Scattered across multiple places means you’ll never find it later
  • Obsidian’s wiki-links connect ideas to each other
  • Search works in one place
  • Backup is simple

So the content seed system went inside BrainFucked. As 10-Blog/seeds/.

What I’ve Built So Far

With this system, here’s what I’ve built:

Hugo blog: Set up with PaperMod theme. English by default, Korean supported. Hosted on Cloudflare Workers at raoulcoutard.com.

7 content seeds: 5 ideas, 1 prompt, 2 misses harvested already. I’m using those seeds to write this very post.

Automation plugin: A plugin that automatically scans previous sessions to harvest seeds when OpenCode starts. Still experimental, but the direction is right.

This Post Is Proof

The interesting thing is that this post itself proves the system works.

  • Saved the original prompt as a seed → quoted it in this post
  • Saved the “one hub” idea as a seed → became a section in this post
  • The process of building the blog → became blog content

Meta-recursive. Building a system to create content, and that process becomes content.

Next Up

It wasn’t perfectly smooth. Symlinks were pointing to wrong paths and I struggled for a while. Bash scripts failed silently for strange reasons. I deliberated over where to store API keys.

In the next post, I’ll share the failures from this process and what I learned from them.

This post was written in conversation with AI, and that conversation itself became a content seed.

Series So Far

  1. Building a Blog with AI: The Beginning
  2. Learning from Failure: The Value of Small Mistakes
  3. Content Begets Content: Discovering Meta-Recursion

More to come…