Essay
From vibe coding to vibe engineering
What changed when I started using Codex seriously, and why engineering judgment still matters.
I had already been using ChatGPT heavily in my personal life before I ever opened Codex.
It had become part of how I thought through decisions, built routines, and stayed accountable to goals that were easy to talk about and much harder to follow through on. One of the biggest examples was health. I used it to help shape a new set of habits and a protocol I could actually stick to, and that process helped me lose 60 pounds.
So when Codex showed up as part of my account, I had to try it.
What surprised me was not that it could write code. Plenty of tools can produce code-shaped output. What surprised me was how quickly it helped me move from idea to working system when paired with real engineering experience.
In less than a week, it helped me go from zero to one on the Watts of Love app. The app was deployed in the cloud on a stack I had never used, with technologies I had never used, and it still felt like progress instead of chaos.
That said, I do not think I could have done the same thing without an engineering background.
There were too many moments where the right move was not obvious from the prompt alone. There were architecture decisions, product tradeoffs, implementation boundaries, naming choices, and moments where speed had to give way to judgment. That is where the experience matters. The best description I have found for it is that I am less interested in vibe coding than I am in vibe engineering.
Vibe coding is letting momentum carry you forward and hoping the code keeps up. Vibe engineering is using that same momentum, but knowing when to slow down, inspect the foundation, and make the call a less experienced builder might miss.
That distinction matters because Codex is most powerful when it is paired with someone who can steer.
What it gives me is leverage.
It helps me explore faster, scaffold faster, debug faster, and close loops that would otherwise sit open too long. It also helps me keep progress moving during the kind of day that usually fragments technical attention. I can be in meetings and still make useful forward motion on work that often gets deprioritized, like moving neglected internal tooling to a supported Node version or cleaning up the kind of operational debt teams rarely have enough time to touch.
That does not mean the human disappears. If anything, it makes taste, judgment, and technical leadership more important.
The value is not in getting rid of the engineer. The value is in extending the range of what one engineer can do in a day without dropping standards.
I think that is what excites me most right now.
Codex has not made engineering judgment less relevant. It has made that judgment more leveraged. It has made it easier to turn intent into momentum. And for someone who still likes to build while leading teams, that is a very compelling place to be.