Vibe Coding? Love the tool, hate the name

I’m crusty and I hate the term “vibe coding.” It smacks of slang from a generation I don't belong to. But since it’s the established industry term, I'll begrudgingly use it.

And I have been using it. A lot.

In the last two weeks, I've put AI-powered development tools to the test and accomplished what would have previously taken me a full year of work. This experience has convinced me of one thing: Human coding is dead. Human-led AI coding is the only path forward.

But don't believe the hype you've read—either for or against it. These tools aren't magic. It’s just that the core skill has changed. It’s not in writing good code. It's in problem analysis, system design, and prompt engineering.

Skill shift: From coder to analyst

I’ve been fortunate to have worked closely with very smart software, ML, and AI engineers over the years, and I’ve picked up a talent for solution architecture and software engineering. I’m not technically skilled in the actual writing of code, but I know how to decompose a problem into an effective algorithm, and I know what standard software design looks like and what to ask for. That all has to come from the human (for the time being), so if you are an aspiring coder, stop learning how to write code, start learning problem analysis, system design, and prompt engineering (focusing on vibe coding).

To give you a sense of what I’ve been able to accomplish in roughly 2 weeks of part time work (by platform).

Case Study: A year of work in two weeks

On Google’s AI Studio

• Blueprint Quote & Project Planner: I built a prototype that reads technical blueprints, drafts a quote based on an AI-driven "take-off," and generates an integrated work plan considering equipment, inventory, and staff skills. It's a rough proof of concept that needs better engineering to become an alpha candidate.

• AI-Assisted Storytelling App for Kids: A more mature prototype that aids children in writing stories. My unique take uses GenAI in an unexpected way, aligned with educational literature to foster creativity and teach AI literacy. This is in limited alpha testing with friends and family.

On Microsoft's VS Code

• Structured Operational Prompting (SOP) Research: This is my core research into a prompt framework for reducing hallucinations in narrow-domain tasks. Early results show quantifiable "lift," and I'm now engaging colleagues for independent validation.

• gnosisPWB (gnosis Prompt Workbench): I realized my research pipeline was a reusable product and spun it off into its own repo with a modular workflow, analytics, and professional Python packaging.

  • A Note on the Code: You can find the repo here. Be warned: I am learning Python and Git as I go. Treat it as a novelty for now, not a production-ready tool.

The Power Curve: Accelerated problem solving with AI

These tools are POWERFUL. This would have represented a full year’s work for me full time, and I was able to deliver all of this in 2 weeks of part time work.

Would you like to know how you can do the same? I’ve started developing a method that I believe will create highly reproducible results for others. It’s only an idea that needs experimental validation, but it’s the beginning of a repeatable method I'm formalizing. If you're building with these tools, follow along and compare notes. We're all figuring this out together.