You’re new to AI. You’re not new to this.

It is an odd thing to have decades of experience and feel like a beginner again. You know how to deliver software. Then you watch a demonstration where someone produces an application in minutes, and for a moment your own way of working feels very far away.

I don’t think the useful response is to dismiss the demonstration. I don’t think it is to dismiss yourself, either.

Separate the unfamiliar from the forgotten

You may be unfamiliar with an AI coding assistant. That does not mean you have forgotten how to investigate an ambiguous requirement, spot an awkward dependency, or ask what happens when a service is unavailable. Those are useful things to bring to a tool that can produce a plausible answer before you have finished considering the question.

Decades in enterprise software teach you to look beyond the screen in front of you. There is usually existing data, another system, an operational constraint, and a person who has to live with the result. A convincing demo does not settle those questions.

Experience needs a little room to move

There is a trap here for those of us who have been around a while. We can turn hard-earned judgment into a reason never to try anything unfamiliar. “I’ve seen this before” can be useful pattern recognition. It can also end a conversation too early.

I want to keep the habit of checking the work while making room for a different way of producing it. That means letting the tool attempt something, inspecting what actually happened, and being willing to change my opinion.

Try this on something you understand

Choose a small function in a project you can safely experiment with. Before using an assistant, write down what it should do, what it must preserve, and two ways it might fail. Ask the assistant to explain the function and propose a change, without editing it yet.

Compare its explanation with your own. Where did it assume something? What did it miss? What did it notice that you did not? Then, if the proposal makes sense, let it make one small change on a branch. Read the diff and run the relevant checks.

This gives you something more useful than a verdict on whether AI is good or bad. It gives you a first sense of where your judgment helps and where the tool helps.

You don’t have to perform enthusiasm

Curiosity is enough to begin. You can be interested and unsettled at the same time. You can enjoy a faster way to work and still care about understanding what you ship.

That is where I want this publication to live: close to the actual work, honest about the uncertainty, and welcoming to people with a lot of experience who still have questions.

Thanks for reading.
— Vince