Does the Developer Job Still Exist in the Age of AI?

I had the pleasure of being a guest of Florian Amstutz on HUMAN, the PeopleUp podcast series devoted to AI and our relationship with work. The episode theme captures the anxiety of the moment well: does your job still exist? We spoke openly about artificial intelligence, the daily life of a freelance developer, and what is really changing in the craft.

Code Has Not Disappeared, It Has Changed in Nature

For a long time, coding assistance failed to deliver. Every three or four months I would test the tools again, and until late 2025 it was never convincing. The context window was too small, the human always had more information than the machine, so it did not really work.

Everything shifted at the start of 2026, with models able to handle far more context, and tools to frame them properly. Concretely, I just shipped an entire application without writing a single line of code myself. I spent my time giving context, explaining precisely what I wanted, setting the boundaries and steering assistants.

I did not write code, but I spent a huge amount of time giving context. The time saved on coding, I reinvested in explaining exactly what I wanted.

The result on that project: at equal time, I delivered far more features to my partner.

"The Problem Is No Longer What to Build, but What to Choose"

This is probably the most important idea of the episode. A feature that used to take two weeks of development can now take an afternoon. Before, out of a list of 200 ideas, you would build three. Today, you can build forty.

Except every feature you add also brings its share of bugs, complexity and decisions. The field of possibilities explodes, and with it the number of trade-offs to make.

If there are ten times more possibilities, there are ten times more decisions to make.

The difficulty has shifted. It is no longer in the execution, it is in the choice.

Why Knowing How to Code Is Still Essential

If this new way of working works for me, it is precisely because I know how to code. The moment an assistant heads in the wrong direction, I see it instantly and can correct course. Expertise does not disappear, it moves toward the ability to steer, judge and decide.

To be honest, this shift suits me well:

I have never been passionate about code itself. I have always been passionate about what you can do with code.

What drives me is the product, understanding the business need, building something useful. That is one of the reasons I went freelance. But I know this is not true for every developer: some are passionate about code itself, and for them the transition will be harder.

AI Does Not Speed Up Everything: the Limit Is Human

We often imagine that AI speeds up everything, infinitely. In reality, the more people you add to a project, the smaller the gain. Knowledge must still be shared, disagreements settled, directions aligned. I lived this in a startup that grew from fewer than ten to thirty-five people: we were three or four times more numerous, without going three or four times faster. As long as there is a human in the loop, this limit remains.

Small Structures, New Opportunities

My conviction for the coming years: we will see many more small businesses and very niche products. Highly specific applications that would never have existed before, because they would never have justified a team and a large investment. A freelancer or a small agency can now tackle that ground.

Large structures will keep existing, because they serve a different market. When I worked for a large group like Nestlé, the stake was not development speed, but the meetings, the reporting, the security and the standards. That need does not change with AI.

What Does Not Change: Relationship and Meaning

Many jobs will be affected, some deeply, like medicine where we already diagnose things that were undetectable a few years ago. But the human side remains: understanding a partner, translating a mission into a concrete solution, bringing judgment.

You can see it clearly with design. AI generates a logo in two seconds, but without the thinking and context understanding that a good designer brings. Their value does not disappear, quite the opposite.

Toni's Advice: Stay Curious

To those whose job is starting to be affected, my advice fits in one word: curiosity. Take an interest in what these tools can do, stay flexible, accept that the craft will change. AI is also a remarkable tool to build skills:

It is the best teacher that has ever existed. You do not understand? Ask it to explain again with a football example, or with apples. It will always find a way.

Optimistic, but Nuanced

On the future, I remain rather optimistic, with nuance. Will there be excesses? Probably, but there already are. As with every industrial revolution, there will be excesses on both sides, neither all good nor all bad. One certainty though: we are going to live through major changes, and we will have to go through them to understand where they lead.

If all of this lets us reclaim time for things that make sense, then it is good news.

Thanks to Florian Amstutz and PeopleUp for the invitation and the quality of the conversation.