Claude and Codex

How I Got Into It

I started using Claude and Codex because I wanted more building leverage, not because AI sounded trendy. At some point it became obvious that a big part of the bottleneck was no longer only what I knew. It was how fast I could turn what I knew into working output.

So I brought them into the process as assistants. That is still the right word for how I think about them.

The Learning Process

The real learning curve was not opening the tools. It was learning how to work with them honestly. How to ask better questions. How to review what they give back. How to know when the answer is useful, when it is sloppy, and when I should ignore it and do the thinking myself.

That part mattered a lot. These tools get better when my judgment gets better.

How I Use It Now

I use them as AI coding and building assistants inside my normal workflow. They help me explore codebases faster, refactor with less friction, debug strange issues, scaffold features, and keep momentum when I am moving across a lot of different layers at once.

They fit most naturally inside VS Code and inside the larger rhythm of AI-Assisted Development. They are especially useful when I am building with things like Next.js, Node.js, and the rest of my stack.

What It Changed

They changed the scale of what a small builder can realistically cover. I still need taste, judgment, and actual understanding, but the distance between idea and execution is much shorter now.

They also made solo or lean-team building feel less narrow. I can test more, move faster, and stay closer to the work without pretending I have a large team behind me.

Related

AI-Assisted Development · VS Code · GitHub · Next.js · Node.js