The year AI threatened our jobs: 2025 review and goals for 2026

What a year 2025 was, right?

Who would have told us at the beginning of the year, when AI was still just a pretty smart copilot, that we would end up watching software developers question the safety of our jobs.

Looking back at 2025

For me it has been a year of many changes (I guess I always say that when I look back at the year ending). But in this case it is very true, because I made drastic professional changes both at the beginning and at the end of the year.

Early year: destroy to rebuild

The year started with a big change: in 2024 I decided to grow the company with two hires.

That jump coincided with a sharp drop in demand for Android training, which made the year quite difficult. My monthly goal was basically to invoice enough to pay salaries and fixed costs.

That forced me to rethink the company goals, which were originally about growing and reaching more developers, but were not aligned with the original reason I decided to start the business.

So I made one of the hardest decisions of my professional life: I let go of the two people who had been with me for years, and I continued on my own.

Moving closer to AI

That gave me a lot of freedom, because I no longer had to worry about high revenue, and I could spend time on what truly fascinated me: Artificial Intelligence.

My first big test was this website. I left WordPress after… 15 years? And built a static site with Astro, which allowed me to learn a lot and have full control of the development.

I know very little about web development, so getting the site to replace my old WordPress was a huge challenge. But I did it, and learned a ton in the process.

And above all, it changed my perception of what was possible with AI.

From that moment I started building tools for my day to day, like a YouTube publisher or a shorts/reels generator.

They were not easy to get working, but with patience the projects came out well and saved me a lot of time. Today, they would literally be a single prompt.

The seed of AI Expert

I had already been fascinated by AI for months, and I talked a lot with Nino (as obsessed as I am) about the future we saw.

He kept telling me we had to build training for this, that people were not able to unlock its full potential. At first I was not fully convinced.

But slowly it took shape. Around March or April we decided that after summer we would launch AI Expert.

The reception was crazy: in September we opened the spots for the November edition, and in less than two weeks the 40 spots were gone.

Today we already filled the 40 spots for January and more than 20 for March.

We are incredibly grateful.

Agentic CLIs

What changed my AI workflow the most was starting to play with Claude Code.

During the summer I decided to give it a real shot. I do not know why, but because you run it from a terminal, it unlocked again what AI could do for us.

Not only that: at that time Claude 4 (I think) was so powerful that AI stopped being a copilot and started acting like the developer.

That is when I realized AI was changing the world, and the way we used it also had to change.

How? That is something probably nobody knows for sure yet.

This has evolved so much that today, with models like Claude Opus 4.5 and GPT-5.2-Codex, AI can achieve results that felt like science fiction at the beginning of the year.

My current stack is Codex-CLI GPT-5.2-Codex High. We will see where this goes.

Burning the ships

For years, people have associated me with Android and Kotlin. All my training and authority were built on that.

But at the end of the year (literally a few weeks ago) I had to make a tough decision.

My Android and Kotlin trainings were getting very outdated, and interest in them keeps falling.

So I had to decide whether to update them (which would take months) or accept that their time had passed.

Also, my head is elsewhere. I am fascinated by AI and the future I see in it. Updating all the training content felt overwhelming.

So I decided to close sales of my Android and Kotlin trainings, and offer them only to companies, where live delivery lets me keep the quality and freshness I always wanted for DevExpert.

The end-of-year shock

I do not know if this is hype or reality, but Opus 4.5 and GPT-5.2 blew through the limits that grounded my beliefs about the future of AI.

I have always defended that programmers will not disappear. I still think that. But what I am starting to see is that the sector is changing (it already is) dramatically.

At levels I did not expect yet.

Writing a simple prompt and having AI deliver a full feature that works on the first try, with that level of quality (functionally and in code), is something you have to experience to internalize.

What we assumed at the beginning of the year that a programmer should do is nowhere near what we will do by late (or even mid) 2026.

It is true that companies do not move as fast as technology, especially in Spain. So I do not want to spread fear that most people will not feel. But I think it is important to stay alert about what is coming. Those who can exploit the gap will benefit from it.

Goals for 2026

To be honest, for years my goals have been quite flexible.

I have accepted that the ideal is to flow with whatever comes, and 2026 will not be an exception.

Full focus on AI

The decision to close the old trainings was so I would not spread myself thin and could focus on what I am most passionate about now: AI.

So in 2026 my goal is to keep studying, researching, and laying the foundations for what AI-powered software development will become.

And of course, apply all of this in AI Expert.

Give more visibility to what I do

Something I am bad at is telling people what I do every day.

This year I have done many interesting things thanks to AI, and even though I have shared some on social media, I think most of you did not even notice.

I want to change that, and share more of my day-to-day, mainly on X, which is the platform that best fits this.

What I try, what fails, what works, what ideas I get - and making most of it open source.

And of course I will keep creating videos on YouTube and publishing on LinkedIn.

Do not put too much pressure on myself

I want to leave space for ideas to flow, to explore, to try and fail, and to learn along the way.

I have a couple of product ideas that, if there is time and motivation, I would like to launch. But I do not want to force it.

And that is it.

I am sure a thousand ideas and opportunities will appear. So I do not want to lock myself into a thousand obligations that will stop me from taking advantage of them.

So, where are we going?

Honestly, I do not know. I do not think anyone does.

What I do believe is that, bubbles aside, software development has changed drastically and is not going back.

As much as it might hurt, or as much as our identity might be based on writing code.

I think few times in programming history we have seen such a drastic change. We can get angry and deny it, or take the lead and keep training, as we always have.

What happens to people starting today? And juniors? Will there be more opportunities or fewer? Or will they be different?

I do not know. Right now we are at a point where learning to program is still essential to guide AI correctly. But in the future, the required skills might be slightly (or very) different.

But I also know that it has never been so easy to go from an idea to something real, and I plan to exploit that by building every crazy thing that crosses my mind.

If it works, great. If it does not, it has never been cheaper to discard code - and I will have learned something.

And I will be here to share my path and, if you want, help you adapt to this new world.