AI Coding, Craft, and Ownership: Which Part of Building Would You Never Delegate?

AI coding tools do more than change how fast we build. They change which part of the work is still ours. A reflection on why the same tools energise some developers and drain others — and where the boundary between leverage and ownership actually sits.

AI coding tools do more than change how fast we build.

They change which part of the work is still ours.

A short post from Marina Wyss made me think about why the same tools energise some developers and drain others.

Two kinds of developers

Some people came to coding because they wanted to turn an idea into something real. For them, agents remove friction. A rough concept can become a working product much faster. That feels like energy.

Others fell in love with the craft itself: understanding a system, chasing a bug, making trade-offs, shaping the code line by line. If AI takes too much of that loop, "more productive" can feel strangely empty.

The same tool. Opposite experiences.

Where I sit

I sit somewhere in the middle.

I use AI heavily because it lets me take an idea much further on my own. But I stay close to architecture, control boundaries, testing, and the decisions I may have to defend later.

That is the boundary I care about. Let the tool carry repetition and open up options. Keep ownership of the judgement.

The real question

Productivity is not the only question.

If AI has made coding less enjoyable for you, it may have removed more than drudgery. It may have removed part of the meaning.

Which part of building would you never delegate?

---

*The post that sparked this was Marina Wyss's YouTube community carousel: https://www.youtube.com/post/Ugkx6CpqnZ85Zj4p_j9Sx4IWVuiJzKbK9IFg*

#AICoding #SoftwareCraft #DeveloperProductivity #AI #Engineering

Topics: AI, AI Coding, Software Engineering, Developer Productivity, Craft, Agentic AI