Do people even read blogs anymore?

The modern internet feels loud, noisy, infinitely algorithmic — short-form everything, subscription-based everything — ultimately optimized for attention, not depth. A long-form engineering blog feels almost… out of step.

And yet, I rebuilt my site as a blog anyway.

Not because I’m chasing traffic. Not because I think it’ll trend on Twitter. And definitely not because I’ve cracked some grand content strategy. I rebuilt it as a very real learning experiment.

The Real Motivation

I wanted to build something from scratch — in the technology stack near and dear to my heart — with intention. More importantly, I wanted to see, for myself, how far AI-assisted development has actually come.

There’s a big difference between reading about AI coding agents and actually handing them meaningful responsibility to build something from concept to deployment. I wanted to test that difference firsthand.

The Experiment

Instead of just prompting for snippets, I structured the workflow deliberately.

I started with bullet-point feature definitions. Then I used the Agent itself to refine and expand my messy notes into a proper Product Requirements Document (PRD), reviewed it, and reformatted it as Markdown. That spec was fed into a coding agent, which generated and scaffolded the implementation.

It wasn’t just:

“Aye yo! Agent — go forth and generate code.”

I had to refine, use, test, and add features one by one, iterating until the result felt usable and acceptable.

But here’s what genuinely surprised me.

At every step along the way, I felt like I had an enormously talented, multi-faceted, tireless partner in building things out — a real ride-or-die coding bredren. This wasn’t autocomplete. It felt like delegating tasks to a very fast, very literal junior engineer who never gets tired and can context-switch instantly.

With structure and clarity, it moves at remarkable speed. I was watching features churn out and get refactored to my liking in minutes.

The Agent didn’t only write code. It:

  • Wrote and refined specifications
  • Reached outside the TUI to use OS-level tools
  • Searched for libraries across multiple sources and suggested viable options
  • Reviewed documentation and implemented required features
  • Scraped content from my old blog and reformatted it for the new system
  • Built and ran the application
  • Curled the locally hosted site and validated output against the feature set
  • Parsed error logs when failures occurred
  • Suggested and executed fixes
  • Re-tested until deployment was working

This is by no means an exhaustive list, but the breadth of capability was eye-opening.

The Important Nuance

AI does not 100% own judgment.

It doesn’t feel the weight of future maintenance. It doesn’t instinctively push back against creeping complexity — though to be fair, it did make several technically sound suggestions when implementations became overly complex. It doesn’t pause and ask, “Will I regret this in six months?”

That’s still our job.

If you have development experience, you can guide the agent toward solid patterns and reasonable structure, and it will produce very strong output in record time. If you don’t guide it carefully, you’ll get something that works — and quietly accumulates technical debt.

Let’s Quickly Touch on Technical Debt

If you look at the code — and not just vibe through it — you’ll see imperfections. There are compromises, plenty of “good enough for now” decisions, and spots where speed was clearly prioritized. This blog will not win architecture awards.

But that’s not entirely the agent’s fault.

The Agent will build exactly what you specify — clearly or unclearly. If your spec is vague, your architecture will be vague. If your constraints are loose, your boundaries will be loose.

The agent is powerful, but direction still matters. Used intentionally, it’s a power tool. Used casually, it builds fast… and sloppy.

That tension is fascinating.

So… Do People Read Blogs Anymore?

Honestly? We’ll see.

I’ll measure, observe, and take note. But here’s what I already know: I built something. I learned a ton. I reactivated engineering muscle memory. I now have firsthand experience with AI agents operating at a level that would have felt like science fiction not long ago.

Maybe that’s the real point.

Sharpening your tools — especially in an AI-augmented world — definitely matters.

If you’re interested, I’m happy to write a deeper technical breakdown of the stack, workflow, and specific agent interactions that made this possible. Let me know.

That part of the story is also interesting… though I’m not sure I’ll have the time.

Or maybe I’ll let my Agent do it 😊