Almost everyone who works in a team knows this. We have a colleague who writes code, earns a decent salary, and no one has a problem with him. But his work—routine Java tasks, repetitive scripts, standard CRUD applications—is being swept into the hands of AI tools like an avalanche. He does not realize it yet, but this is happening right now.
The comfortable developer middle class is disappearing
Just a few years ago, it was enough to be a solid programmer, know a few frameworks, and job offers came naturally. A salary of around CZK 150,000 per month, a stable company, no major stress. That era is over. The comfortable middle tier of developers is falling apart—and AI is the catalyst that has dramatically accelerated the entire process.
Researchers Dr. Matt Beane, Jonathan Hassell, and Steve Yegge described it precisely: job roles are “merging and splitting like nuclear reactions.” The market is polarizing. You are either at the top or at the bottom. The middle is collapsing. And yet—and this is the paradoxical part—NVIDIA is currently looking for more than 700 software engineers, compared with just nine AI researchers. OpenAI has four times as many open positions for engineers as it does for researchers. Programming is not dying. It is changing beyond recognition.
Three developer castes: Where do you belong?
The IT job market has split into three tiers. Each offers a different fate.
The top tier commands salaries ranging from CZK 250,000 to 500,000 per month and above. These people design the architectures of complex AI systems, connect technical infrastructure with business strategy, and turn chaotic prototypes into products for millions of users. They are known as “apex builders.”
The hybrid middle ranges between CZK 150,000 and 300,000. This is where developers who have combined technical expertise with product thinking, design, or operations are found. For example, a “fleet supervisor”—someone who manages dozens of AI agents like an air traffic controller. Or an “agent expert”—a tax lawyer who has learned prompt engineering and monitors when AI hallucinates about depreciation schedules.
The automatable tail—and this is where it hurts—includes routine, repetitive coding tasks paying CZK 80,000 to 130,000. This group faces pressure from two directions: AI can handle the same tasks 55% faster, while global competitors offer the same work for a fraction of the price. This is not a temporary fluctuation, but a structural change.
Cursor and similar tools really do speed up developers’ work by more than half. That sounds great, doesn’t it? But completing routine tasks faster does not create more routine jobs—on the contrary, it eliminates the economic rationale for paying humans to do them.
Meanwhile, the U.S. Bureau of Labor Statistics predicts 17% job growth for software developers by 2032. How do these facts fit together? Demand is not declining—it is simply shifting upward in terms of quality. AI is intensifying pressures that have always existed: efficiency, global competition, and cost savings. It is just that everything is now moving much faster.
New hybrid roles that do not yet have names
Positions that did not even have names two years ago are beginning to appear on job portals.
Platform Designer—one-third product manager, one-third designer, one-third systems engineer. Their task? To make chaotic AI features appear as reliable as a calculator to users.
Fleet Fixer—debugs “conversations between machines.” Someone compared it to family therapy for robots. It requires a deep understanding of how AI agents behave when they start arguing.
AI Quality Assurance Engineer—their value lies in what they reject. They ensure that AI does not generate “nonsense” that passes automated tests.
The common denominator of all these roles? Deep specialization in one field combined with practical skills in adjacent disciplines. Seemingly unrelated expertise is becoming a superpower.
What now? Guidance based on age and circumstances
Are you a recent graduate? Choose your path deliberately. The top tier means 60 or more hours a week, intense pressure, and enormous impact. The hybrid middle offers meaningful work that funds a life outside the office. Both are legitimate choices—just make the decision yourself rather than leaving it to chance.
Are you an experienced developer with fifteen years of practice? Your experience has zero value if it is frozen in time. Everything you know becomes context for AI—not a substitute for it. Stop doing routine work and start managing it. Become a “force multiplier”—someone who multiplies the performance of others.
And one thing applies to everyone: honestly assess where your work falls. If it can be described as “repetitive coding tasks,” you have roughly 12 to 18 months to move up before the pressure on compensation fully intensifies. The market will not wait for you to decide. The question is not whether AI will change your career—but whether you will change it yourself before AI does it for you.
Sources: systemonline.cz and itrevolution.com
