Last month, I wrote about my search for joy in a new world of programming. I realized this month: I’d fallen into the trap of treating my own work too much as producing code, fixing bugs, and assembling commoditized software. But that’s not what I advise others to do, so why am I assessing myself like that then? It's like treating a carpenter, because since IKEA furniture is a thing, his job has become easier. It's just not how it works.

There is definitely some truth in my last post, as well as in the IKEA metaphor. Code on its own becomes less valuable. And yes, code has always been one of the fun parts of the engineering process, or at least it felt like that, but it's not the use case itself.

If my last post was about losing the plot, this is about finding it again. By embracing what we should be doing and wiping away some mistakes. 

We all agree that the art of software development was never about the code itself. It was about thinking the puzzle, the optimisation, the moment when a problem clicks into place, and you realise you’ve built something that works. I also realized last month that joy wasn’t in the syntax only; it is as much in the research, trying out new techniques and in the solving itself. It's in the journey and the craft of engineering.

With AI too many people are currently only talking about producing code. Especially the ones that say agents are going to replace developers within the next few days. Or the ones that became programmers themselves since vibecoding. It’s almost like a new myth showed up, or at least that's how it felt at my side, developers do produce code. And because of that, everyone is now a developer.

But we all know developing is so much more than that. It is as much in the research, trying out new techniques, optimisation, and in the solving itself. Don't let yourself be reduced to an assembly-line worker who produces code for commoditised software.

I realised, AI didn’t take my joy; I gave it away by falling into that trap and by mistaking the tool for the craft. I tried to use it to outsource my thinking. And that's the worst way to use AI for. Don't fall into the same trap when assessing or using AI. Start using it to develop sharper thinking, and force yourself to let your brain work.

Great articles and podcasts that inspired me on this are: "Why developers are sleepwalking", which is an episode of The Agentic Review. And an article in ct magazine, where researcher Michael Gerlich explains; When you trust on gen AI, then you disable critical thinking. He explains very well that structured processes can yield optimal results and can even promote your personal development. That's not a threat. That's a liberation and definitely where we can find joy again.

So, after realising all this. What shift did my day-to-day joy make?

I reject the idea that my role, when using AI, is to be a person throwing tons of documents to a robot to do the work and the thinking for me. Structured processes and tools forced me to be a captain again.

That's where frameworks like GSD, Superpowers and Caveman came around the corner. When using these, I realised it's not the IKEA cabinet I need to screw together; it's the IKEA cabinet I need to research, optimise, or even invent. It's not about big, all-you-can-eat large language models; it's about optimising the language you talk to it (Caveman), structuring inputs and outputs (GSD, Superpowers), and developing models to use available resources wisely (Mellum). There is still so much to discover, optimise, and invent within this landscape. And for all of this, real craftsmanship is still needed.

Another joyful fact: I realised the direction I stepped into with Backlot has great potential, thanks to its optimised, structured way of working, which we call Movie Pattern Practice. This feels like a smart and natural fit for a joyful future for me as a framework developer and someone who likes to use and train his own brain every day.