It has been a few months since I truly sat down and wrote code end-to-end. At first I was skeptical about letting LLMs make changes. But now, with agents like Claude, because I understand what is happening under the hood, work feels different. It almost feels like a bank job. We supervise. We decide direction. Systems execute.
This shift will multiply productivity. But it also made me realise something deeper. We are becoming increasingly dependent on systems we do not fully control.
You can see this in everyday life. Most of us no longer remember routes. We rely on GPS. We no longer memorise phone numbers or small details because everything is stored in our devices and cloud accounts.
Convenience works so well that we assume it will always be available. Right from our coffee ground and distribution availability, to electricity to anything.
AI is following the same pattern. Very few still build or self-host, but for most of us, access replaces ownership.
It is both good and bad. It is the evolution. But it also quietly changes what we practise and what we begin to call a skill.
It also matters that much of this infrastructure is controlled by a small number of powerful players. When tools work smoothly, we rarely question alternatives or our ability to function without them.
At the same time, these systems expand what we can do. If machines handle repetition and even parts of "routine" thinking, we get more room to focus on deeper problems, explore new ideas, and build things we might not have attempted earlier.
This is why staying close to fundamentals still feels important. Not as a backup plan, but as a way to remain grounded. When execution becomes easier, understanding becomes more valuable. It helps us question, shape ideas better, and go beyond what tools suggest.
So maybe the real shift is not from coding to not coding. It is from doing everything ourselves to being more intentional about what we choose to learn, build, and rely on next.