The invisible shift

4 minutes of reading

Public debate mainly focuses on large companies announcing thousands of job cuts. Yet a much more diffuse transformation could be taking place within small businesses.

In Switzerland, SMEs account for more than 99% of companies and around two-thirds of employment.[1] Every day, they purchase services in accounting, communication, web development, writing, translation, administrative support and consulting.

AI now allows an entrepreneur to carry out part of these activities themselves or automate their production. They do not need to completely replace their accountant or agency. It is enough to buy two hours of review where they previously bought twenty hours of production.

This reduction seems insignificant when looking at a single company. Across hundreds of thousands of individual decisions, however, it could lead to a major contraction in service markets.

A transformation that is difficult to observe

Early studies of freelance work are already detecting this phenomenon.

A study based on a large online labour platform found that, after the arrival of ChatGPT, freelancers working in activities most exposed to generative AI experienced a decline in both activity and income.[2]

A study published in 2026, based on actual payments made by US companies between 2021 and 2025, also found a shift in spending away from freelance platforms and towards AI model providers. Among the companies most exposed, every dollar less spent on outsourced labour was associated with around three cents of additional spending on AI.[3]

This shift may remain invisible for some time.

It does not necessarily appear in the form of layoffs. It shows up through contracts that are not renewed, falling rates, underemployed freelancers, agencies that do not replace people who leave, and young people who never find the entry-level positions that were previously necessary to learn the job.

The disappearance of the junior ladder

One of the most important risks concerns junior roles.

A company may still need an expert to review a result, take responsibility or handle exceptions. It will need far fewer people to prepare documents, enter information, produce first drafts or carry out routine operations.

Yet these tasks have traditionally been how people learn a profession.

If they are automated, companies will need to find new ways to train future experts. Without this reconstruction, they could reduce their immediate costs while gradually destroying their own pipeline of skills.

The International Labour Organization finds that clerical occupations remain among the most exposed to generative AI, while exposure is also increasing in some technical and professional occupations.[4]

Exposure does not automatically mean that a profession will disappear. A profession usually combines different tasks, some of which can be automated and others which cannot.

This also raises a generational issue. Part of the younger generation may enter the labour market at the exact moment when traditional entry points are becoming narrower.

However, it is important not to mistake the most visible minorities for an entire generation. Platforms tend to promote content that generates a reaction. Aggressive, highly ideological or constantly outraged voices can therefore appear far more prominent than people who simply live, work and develop without continuously producing ideological content.

A transformation without layoffs

The most visible manifestation of AI may be the arrival of highly capable machines or the layoffs announced by large companies.

But its deepest economic impact could be much less visible: millions of small decisions through which businesses buy fewer hours, produce more themselves and concentrate their spending on a smaller number of people able to design and supervise systems.

The key question may therefore no longer be only which jobs will disappear. It will also be how value and work are redistributed.

References