Stepping Back from the AI Productivity Race

In my last post [1], I discussed how 2025 was shaped by the new possibilities of AI-assisted programming. It has been a wild ride, and like many of you, I jumped headfirst into the world of “Vibe Coding.” However, after several months of riding that wave, I have realized that this euphoria comes with a significant downside. I suspect many of you might feel the same.

The Addiction of Vibe Coding

Vibe Coding feels almost like an addiction. I often find myself stuck in a loop: prompt, wait, test, repeat. AI can generate solutions in minutes that would have previously taken days, depending on one’s experience and routine. It is an incredible rush to see complex logic manifest nearly instantly. You soon realize you can optimize this loop even further by using external tools like task managers or MCP servers, making the process more efficient than ever before.

However, I noticed a pattern. I was spending significant time and resources prompting scripts and programs that were impressive in the moment but completely forgotten by the next day. I was producing simply because I could, not because I should. Even the “good” solutions often carried a hidden tax: they created more assets to manage, maintain, and worry about than I had before.

When “Better” is the Enemy of “Good”

Someone once told me, “Better is the evil brother of Good.” It is a twist on the old adage that “the better is the enemy of the good,” and I think it hits the nail on the head. With AI at my fingertips, I am constantly tempted to make things “just a little bit better” or to create things I don’t really need.

This is where the Pareto principle, or the 80/20 rule, kicks in. That final 20% of features or polish usually consumes the vast majority of the effort. In a world where AI makes the first 80% feel effortless, we end up wasting our human time chasing that final 20% on projects that might not even matter in the long run. And let’s face it, the last 20% is not the LLM’s strength; by that point, you have likely exceeded the context window twice over.

Finding Real Value in the Noise

Over the last few weeks, I have taken a deliberate step back. Now, before I start a new project or prompt a new tool, I ask myself: “How does this actually make my life easier?”

I want to focus on projects that offer genuine value rather than just stacking new functions that I don’t have the time to use. Ultimately, I want my tech projects to remain a hobby, not a full-time IT operation that I have to babysit. This mindset shift is exactly what led me to simplify my setup, resulting in my recent Service Hub [2] and my streamlined Hugo workflow [3].

My Resolution: Less is More

Looking back at the past year, my biggest takeaway is a classic: Less is more. AI in 2025 felt like a frantic race to see who could adopt new tools and techniques the fastest. But I have realized that most of the code I write today will likely be obsolete or handled automatically by a new AI tool in the near future anyway.

Moving forward, I am shifting my focus. I want to develop solutions that ultimately allow me to spend less time at the computer. I want to free up space for non-technical activities—the things that have been sitting on my to-do list for far too long.

If a tool doesn’t buy back my time or simplify my life, I am learning to let it go. It is time to stop racing and start building with intention.

Sources

[1] My Personal AI Recap of 2025 – Part 1
[2] Building a Service Hub: One Tool, Three Ways to Use It
[3] My Hugo Workflow: From Markdown to Live Post