I Keep Abandoning Blogs — This Time I'm Restarting With AI Alongside

Published:

Honestly, this isn’t my first blog. I’ve started several before, overthought my way into paralysis, and quietly let each one die. So the very first thing I decided this time wasn’t the topic — it was how to keep it alive.

This time’s bet: pair with AI

The reason I always quit is the same: I think too much before I write. I plan the structure, agonize over the niche, fiddle with the design — and the excitement fades before a single post ships.

This time I’m fixing that with a system. I’m making Claude Code my “running mate,” so I act a little before I think. A post at 80% beats a perfect post that never ships. Don’t let drafts sit. That’s the one and only top rule of this blog.

The foundation itself was already AI-paired

“Pairing with AI” is abstract, so here’s the first concrete instance: I had Claude Code set up this very blog.

The only thing I typed to kick it off was this:

I want to launch my personal blog. I work across engineering, data science, project management, and management, so I want content related to that, plus automation like Claude Code, with meta-analysis as a flavor. What should I decide first?

From there it laid out “what to decide first” as a set of options, and I just picked from them. Before I knew it, the Astro scaffold was sitting there.

This is usually exactly where I trip. I start fiddling — framework choices, design — and let the blog die before a single post ships. This time I threw out the question before thinking and just chose from the answers. I cleared the first obstacle the same way I write the posts: alongside AI. Small, but I suspect it’s the part that’s actually working.

What I’ll write: hands-on logs of Claude Code and automation

The main theme is Claude Code and automation — not heavy theory, but whatever I actually did, written up as-is:

I’ll keep the code and steps concrete enough that you can copy them.

Why automation (the honest version)

I currently work across management, data science, engineering, and PM — and of those four, the one most likely to get me promoted is management. But honestly? Management is the one I’m least interested in.

What I actually want is to pair with AI, automate my own work, and break toward something more autonomous — financially, too, eventually. I’ll narrate that process without the polish. Not hiding the honest parts is itself the content here.

Where it grows

I’m not locking in a grand niche up front. As the Claude Code logs pile up, this will probably drift naturally toward “how we work” and “meta-analysis” — looking at my own and others’ experience as data. When that time comes, I’ll let it grow.

One post first

So — this is post number one. I’m already practicing “ship at 80%.”

Next up: how I wired this bilingual setup with Claude Code. Here we go.

← All posts