Not so long ago, explaining what I did for a living was fairly simple.
I'd say I was a developer.
The term was imperfect. Behind it could hide very different realities: web development, backend, frontend, integration, architecture, technical expertise...
But at least we shared roughly the same vocabulary.
Today, I look at certain job postings and sometimes feel like I'm discovering a different profession every week.
AI Engineer.
Applied AI Engineer.
AI Integrator.
AI Integrations Engineer.
AI Solutions Architect.
AI Consultant.
AI Adoption Consultant.
AI Transformation Consultant.
Engagement Manager, AI Implementations.
And probably a few more by the time I finish writing this.
It's not just a matter of vocabulary
One could dismiss this proliferation of titles as a simple HR fad.
Tacking "AI" onto an existing role to make it sound more attractive.
That would be too easy.
Because behind some of these titles, real needs exist.
Models need to be integrated into information systems. Agents need to be built. APIs need to be connected. Data needs to be leveraged. Architectures need to be set up. Processes need to be automated. Usage needs to be secured. Users need support. Results need to be measured.
The work is real.
What seems far less settled is how that work gets carved up into jobs.
Going through the offers I receive, I've started noticing similar responsibilities hiding behind different titles.
Conversely, two companies using the same term can sometimes seem to be looking for noticeably different profiles.
Even LinkedIn, around a single "AI Integrator" alert, suggests postings as different as AI Integrations Engineer, Applied AI Engineer, or Engagement Manager, AI Implementations.
So the market seems to know it needs new skills.
It seems far less certain about which jobs to make out of them.
A developer, but what exactly now?
This is where the phenomenon stops being theoretical for me.
I come from development.
Over time, like many experienced developers, my activity has largely outgrown writing code.
Architecture, APIs, systems integration, infrastructure, understanding needs, technical choices, automation, project support...
Then generative AI arrived.
And it didn't simply add one more technology to the toolbox.
It started shifting the boundaries.
Part of what we used to call development is becoming automatable.
Other skills are gaining importance.
New problems are appearing.
And with them, new titles.
Which raises a fairly uncomfortable question:
how am I supposed to evolve my job when the market itself still seems to be searching for how to name this evolution?
Get trained, sure. But in what, exactly?
The usual answer to a technological shift is fairly obvious: get trained.
Fine.
But for which job?
AI Engineer?
AI Integrator?
AI Architect?
AI Consultant?
Adoption specialist?
And trained in what?
Development with LLMs?
RAG?
Agents?
MCP?
Orchestration?
AI cloud?
Governance?
Change management?
Training programs and certifications obviously exist.
They're trying, in their own way, to answer the same question I am: where does an "AI job" begin and end?
France Compétences, the French national body for skills certification, is working precisely on this difficulty. In its work on emerging or fast-evolving occupations, it distinguishes a genuinely new job from one born of hybridization between several jobs, a full recomposition of its activities and skills, or an existing job whose content is changing significantly. The AI project manager was already listed among its emerging-job examples back in 2020.
This France Compétences framework puts fairly precise words on the problem: before even certifying new skills, one first has to determine whether they amount to a genuinely new job, an evolution of an existing one, or simply complementary skills.
But that's exactly where the gap becomes interesting.
While training organizations and institutions try to formalize these shifts, companies are already experimenting with other ways of carving up the work.
The market moves forward.
Titles appear.
Skills get bundled together.
Then the frameworks try to catch up with what's already emerging.
New jobs... really?
Looking at certain job ads, another hypothesis emerges.
What if some of these "new AI jobs" aren't really new jobs at all?
An AI Integrator often looks a lot like a systems integrator now expected to master models, agents, and automation tools.
An AI Solutions Architect largely remains a solutions architect with new technological building blocks added on.
An AI Adoption Consultant borrows heavily from digital transformation and change management.
An Applied AI Engineer blends software development, architecture, integration, and new skills tied to generative models.
In other words:
AI may not only be creating new jobs. It may be brutally recomposing the ones that already existed.
This is no longer just a hunch. In 2025, the International Labour Organization estimated that one in four workers worldwide held a job with some degree of exposure to generative AI. Its key conclusion here: job transformation is judged more likely than job elimination, largely because human involvement remains necessary in most occupations.
The OECD reaches a complementary observation: most workers exposed to AI won't need to become machine learning or NLP specialists. What's changing is the tasks they perform — and therefore the combinations of skills expected of them.
And we then try to put names on these recompositions.
We're navigating while the map is still being drawn
This is probably what bothers me most about the current situation.
Not that jobs are changing.
They always have.
Not that I have to keep learning.
That's been part of being a developer forever.
The shift is already visible here too. In its 2026 study on digital jobs, France Travail (the French national employment agency) reports that nearly 40% of surveyed establishments cite programming, code production, maintaining IT infrastructure, or learning new languages among the skills that need to evolve. The study adds that AI is expected to assist developers with coding and to become integrated into testing and maintenance.
In other words, even the job I thought I knew how to name is already on the move.
What's different today is the speed.
The technology moves forward.
Companies experiment.
Vendors launch their platforms.
Training organizations build their curricula.
Recruiters invent titles.
Professionals look for the skills they need to acquire.
And all of this happens almost simultaneously.
So many of us have to decide today how to steer several years of a career, based on a market that hasn't yet finished defining the jobs we're supposedly preparing for.
It's a fairly paradoxical situation.
We're told to adapt quickly to AI.
But adapt to what, exactly?
The stammering might be the real signal
I could easily poke fun at this pile-up of new job titles.
It would be tempting.
But probably unfair.
This apparent stammering may be telling us something far more important.
We're not facing a technology that simply replaces a previous one.
We're witnessing a recomposition happening fast enough that the labor market hasn't yet had time to build its vocabulary.
That may actually be the best way to read this period: not as a disorderly succession of new jobs, but as a market trying to name functions while their tasks and boundaries are still in motion. France Compétences talks about hybridization and recomposition. The ILO talks more about transformation than replacement. The OECD observes that AI is reshaping tasks and skills well beyond AI specialists alone. The words differ, but the phenomenon they describe is remarkably close.
Companies increasingly know what they want to do with AI.
They're still figuring out how to organize the work needed to get there.
Jobs will probably settle down eventually.
Some titles will disappear.
Others will become mundane.
Frameworks will emerge.
Training programs will eventually align.
This recomposition matches my own path too. Originally a developer, I now work across a scope that combines architecture, systems integration, infrastructure, automation, and AI agents. So I'm not just looking for the new title that would replace "developer." I'm looking for the accurate way to describe a function I already perform: turning a business intention into an operational system, then organizing the cooperation between its technical, human, and now intelligent components.
But for those going through this transition today, the situation remains far less comfortable:
we have to learn to navigate while the map is still being drawn.
Going further
This article starts from a personal experience of the labor market. Several studies help place that observation within a broader phenomenon of job transformation and recomposition driven by AI.