AI is not just reshaping jobs, it is changing which workers have the advantage.
The old startup playbook rewarded teams that could stack up young, cheap engineers and let them learn by doing. Bloomberg's latest reporting suggests that logic is starting to crack, because AI is absorbing the kind of routine work that once gave junior hires a place to start, and that leaves companies valuing people who can supervise, verify, and decide what AI should do next.
That is a meaningful shift for founders. The core problem is not simply that AI can write code, summarize documents, or sort through data faster than a new graduate. It is that those tasks were also the training ground, the low-risk work that taught people how products are built, how clients behave, and how systems fail. When that pipeline shrinks, the career ladder gets narrower at the bottom and more expensive at the top.
Bloomberg reported on May 15 that several US occupations exposed to AI, including customer service representatives, certain secretaries, and salespeople, posted heavy job losses for a second straight year in 2025. The outlet said an 18-occupation group flagged by the Bureau of Labor Statistics, representing about 10 million jobs, fell 0.2% between May 2024 and May 2025, even as overall employment rose 0.8%. That matters because it shows the pressure is no longer theoretical, it is showing up in labor data now.
The new bias is easy to understand. AI is good at single tasks, but it is much weaker at coordination, context, and accountability. Those are the areas where experienced workers earn their keep, because they have enough domain judgment to know when output is wrong, incomplete, or simply not worth shipping.
That is why older workers may gain leverage rather than lose it. They can do what AI cannot, connect the dots across functions, spot the edge cases, and make the final call when the stakes are real. In practical terms, a company may need fewer entry-level employees who produce first drafts and more seasoned operators who can direct a smaller team, challenge the machine, and keep mistakes from compounding.
This is also why the traditional hiring orthodoxy is under pressure. For years, startups justified junior-heavy teams with the argument that raw talent was cheaper and could scale quickly if given enough responsibility. Bloomberg's reporting, along with recent labor-market coverage from the same newsroom, points to a different reality, one in which AI is shrinking the set of tasks that once made young hires cost-effective in the first place.
There is another layer here that founders should not miss. If AI can complete the basics, then the hiring bar rises, and compensation follows. The market may start paying for judgment instead of throughput. That changes how startups budget, because the incremental cost of one strong senior who can manage AI output may be lower than a larger bench of juniors who still need oversight.
What startups should change
For early-stage companies, the obvious lesson is not to stop hiring junior talent altogether. It is to stop assuming a junior-heavy pyramid is the default growth model. If AI takes on the repetitive work, then the better structure may be a smaller team anchored by senior product, engineering, and operations leaders who can use AI as an amplifier rather than a replacement for management.
That shift has direct consequences for burn rates. A startup that hires fewer low-cost juniors and more expensive experienced people may look less efficient on paper, but it could move faster if those seniors can ship better decisions with less supervision. In other words, the classic startup tradeoff is changing from headcount to judgment density.
The training path also gets more complicated. If the entry-level role no longer provides a wide enough set of tasks to build skill, startups will need new ways to develop talent. That could mean tighter apprenticeship models, more deliberate rotation across functions, or AI-assisted workflows that still expose younger workers to the reasoning behind the work, not just the output.
Bloomberg's recent coverage of the job market makes clear this is not a narrow tech story. It is a labor-market story with startup consequences. The companies that adapt early may discover that AI does not eliminate the need for people, it changes which people matter most and what kind of work deserves a salary premium.
For founders, the useful question is no longer how many juniors can we afford. It is how many people on this team can actually direct the machine, catch its mistakes, and turn its speed into better decisions. That is a different company design, and it is coming fast.
Also read: New York Fed data says AI is not driving the hiring slowdown • Nous Research's new training method could change the economics of LLMs • Microsoft's AI chief says white-collar work may be automated fast
