I Got My Excitement Back (Thanks to AI)

For years the gap between “I have an idea” and “I see it work” stayed the same: wide. A few months ago it got a lot shorter, and something I had not felt in a while came back with it: the urge to build personal projects (on top of the ones I already do every day at work).

The drawer full of ideas

I have a stack of ideas set aside. They are not necessarily projects with a business behind them: more often they are simple ideas, things that solve something useful for me and that might turn out useful to others too. When I see something missing, or that could be done better, I get the urge to fix it and do it my way.

A couple of examples: an app to keep score during padel matches straight from the smartwatch (I know they already exist, but I would want my own version, born to put an end to the inevitable arguments over the score during matches with friends: my sanity was at stake), or an app to catalog the books I read on Kindle.

I never got them off the ground, not because I did not know how to make them, but because I knew the price: quite a few weekends burned and no guarantee of ever reaching something even remotely visible. Just thinking about it was enough to drain the excitement before I even opened the IDE.

There was a second obstacle too. Like in many other professions, my software background does not cover every aspect I would need: to move even a POC or an MVP forward I was missing pieces that fell outside my own skill set, like electronics, frontend or app development. Learning a whole skill from scratch just to validate an idea was an investment that almost never paid off.

What changed

AI and coding agents removed exactly that friction. It is not that they “write everything for you”: it is that the trip from idea to first working prototype went from several Saturdays to a single evening. Seeing something run quickly changes the game, because excitement feeds on results, not on good intentions.

The cost of trying also dropped. Before, every idea was a bet worth weeks, and the bar to even begin was sky-high. Today I can spin up a prototype in a few hours: if it does not work I delete it without much regret, and that pushes me to start ideas I would have shelved.

And then there is the skills gap. The gaps that used to block me I now fill just enough to reach something working, even in areas I do not master: AI has my back, without having to become an expert in everything before I start.

Superpowers, with an asterisk

For people with experience these tools are superpowers: you know what to ask, you spot a wrong answer at a glance, you know where to dig when something does not add up. The judgment is already there, AI just boosts it.

And it is not just speed. Coding agents come up with high-level, sometimes complex technical solutions I would not even have thought of. And bugs, when well documented, they fix in no time, saving me from losing entire days hunting down an error. The one caveat: you have to know exactly what to ask and leave no room for ambiguity, otherwise the help backfires.

There is also something I did not expect: these tools are a learning resource. Watching them produce a result and understanding what they are doing, even where I start from zero, gets me at least to the level of “I know what should be done conceptually”. From there, one step at a time, the skill actually builds up. On one condition, though: you need solid foundations to hang what you learn on, otherwise it all stays on the surface.

The problem is for those starting now. If you skip the struggle, you skip the learning too: that struggle was how you built the judgment that today lets me use these tools well. The risk is a generation that can make code appear without being able to judge it.

Conclusions

The way I see it, it is simple: the excitement is back, and the ideas in the drawer are finally coming out one after another. I do not see these tools as a replacement, but as a multiplier of what I already know how to do.

I got into this line of work for one thing in particular: the pleasure of seeing something I built actually work. At the start I was all about the technical side, about how things were done in detail; with time and experience I shifted toward the value of what I build, rather than how it is built. And that is exactly what my excitement feeds on.

The way I work has changed too: I am no longer the one writing the code, line by line. Now I open many tasks in parallel and spend my time checking, steering and correcting the various agents doing the work in my place. I have made peace with “losing control” over the how, letting AI take the wheel: the destination, though, is still mine to choose, and reaching it quickly gives me enormous satisfaction.

The question I am left with is about those who come next: if the struggle that used to build judgment can be skipped, how will they build theirs? I do not have the answer, and I am wary of anyone who already does.

For now I am enjoying the good part: building things is fun again. The rest I will figure out along the way.