Application Flood 2026: When Everyone Applies with AI — How to Reach a Human in Switzerland

Anyone clicking on a jobs.ch ad in summer 2026 is competing with far more people than a year ago — and a large share of them are using the same assistant you are. The application flood is no longer a claim made by recruiting blogs; it is measurable: more applications for fewer ads, a large share of them AI-assisted, and on the other side HR departments admitting they can barely judge motivation and authenticity any more. This article sorts out which parts of that affect you — and what you do differently from the other 300 in the stack.

What the numbers really say

Three data sets from this year are enough to draw the picture — all from Switzerland, none from a LinkedIn post.

More applications, fewer jobs. JobCloud (jobs.ch, jobup.ch) reported 25 percent more applications started per ad for the first quarter of 2026, with 11.5 percent fewer ads. The second quarter got sharper: 36 percent more in April, almost 50 percent more in May, 22 percent more in June — with around 28 percent fewer active ads than a year earlier. That is the squeeze you feel when applying: more people pushing for every job, and fewer jobs to fall back on. The SECO figures for July fit the picture: 139,276 unemployed, 7.8 percent more than a year ago, a rate of 3.0 percent.

Half of them are AI-assisted — roughly. In spring 2026 the labour-market observatory AMOSA surveyed almost 6,000 job seekers and around 700 companies in 13 cantons. Around two thirds of job seekers have already used AI for their job search, 34 percent regularly, and the most common purpose, at 61 percent, is creating and optimising application documents. Companies estimate that around 39 percent of incoming applications were produced at least partly with AI. Among specialists and managers the share is higher: in its Talent Trends 2026 (1,220 respondents in Switzerland) Michael Page arrives at 86 percent of job seekers using AI.

And the employers? This is where it gets interesting, because the numbers tell two stories. According to AMOSA, only 13 percent of surveyed employers use AI in recruiting so far, almost a quarter among large companies — and only 3.6 percent for pre-screening applications. Michael Page, by contrast, counts 69 percent of employers using AI in the recruitment process. The difference is not a contradiction but the sample: AMOSA surveys broadly, including SMEs in the RAV regions; Michael Page surveys its own clientele, which is larger and more international than average. For you that means: at a corporation or an international firm, expect a system between you and the human. At an SME, a person is probably reading — and that person is currently reading 300 letters that all sound the same.

The real opponent is not the ATS but the sameness

Robert Walters describes the problem from the recruiter's side: some receive over 800 applications for management positions, more than twice as many as the year before — and "many applications look almost identical". That is a global observation, not a Swiss statistic, but the AMOSA figures confirm the direction: 55 percent of surveyed companies say they can recognise AI applications by formal or linguistic features, and more than half say it is becoming increasingly difficult to judge motivation and authenticity.

That "recognising" is less detective work than pattern recognition. Gery Bruederlin of the FHNW says well-prompted texts are almost impossible to identify as AI-generated today — what recruiters recognise is the badly prompted majority: interchangeable openings ("I read your advertisement with great interest"), lists of qualities without evidence, the second paragraph that paraphrases the ad, and in German-speaking Switzerland the German ß, a letter no Swiss word processor produces. How to avoid these patterns is covered in detail in the article on ChatGPT cover letters in Switzerland; here it is about the consequence for your strategy.

Because the second half of the AMOSA figure is the more important one: regular AI users receive around 48 percent more invitations to interviews. Using AI is no flaw. The disadvantage only arises when the application consists only of AI — when there is nothing in it the model could not have written for the other 299 as well. Raiffeisen puts it this way in a guide: AI-generated documents no longer offer an advantage, and an obviously generated application may even be rated negatively. Only 39 percent of employers at Michael Page say AI-assisted applications make them more likely to hire. The rest are neutral to sceptical.

What the Data Protection Act requires for automated pre-screening

Switzerland has no AI law, and the EU AI Act, which classifies recruiting systems as high-risk, does not apply directly to Swiss employers — its high-risk obligations were in any case postponed to 2 December 2027 in July 2026. What applies in Switzerland is the revised Data Protection Act, and Art. 21 FADP contains something very few applicants know:

"The controller shall inform the data subject of a decision that is based exclusively on automated processing and that has legal consequences for the data subject or significantly affects them (automated individual decision)."

And in paragraph 2: the data subject may state their point of view and demand that the decision be reviewed by a natural person. Add to that the right of access in Art. 25: on request, the employer must tell you whether an automated individual decision has been made — and "the logic on which the decision is based".

Where exactly the line runs is something the Federal Data Protection Commissioner has not yet specified for application systems. The legal opinion, for instance from Facincani on arbeitsrecht-aktuell.ch: a ranking that a human then actually reviews and decides on does not fall under Art. 21. A rejection the system sends on its own based purely on a score, with nobody looking, very probably does. In practice, that means for you:

But do not assume the system is the norm. SBB, for instance, writes in its FAQ for applicants that every incoming application is reviewed personally by a recruiting partner; Adecco Switzerland says an algorithm will never decide alone. The bottleneck is usually not the machine. It is the person with 300 open tabs.

Five things that stand out in the stack

When the competition consists of more applications that all sound alike, the winner is not whoever sends even more. The winner is whoever delivers something per application that the others cannot.

1. The CV speaks the language of the ad — literally. Whether a human or a system pre-sorts, both look for the terms from the job description. "Head of accounts receivable" is not "responsible for receivables". Before you write anything, match your CV against the ad and name the three requirements you meet most strongly, using exactly the words that appear there. That takes 20 minutes and is the one part AI does truly reliably for you.

2. One sentence in the letter that only you could write. Not "your innovative corporate culture", but: the company project you know from the annual report or the local paper, the link to your last job, the number you moved there. A single concrete detail lifts a letter out of the pattern. Two are better. The model can build them in — you have to supply them.

3. Fewer applications, each one complete. With 50 percent more competition per ad, the half-hearted application is worthless — it only raises the number of rejections you have to process. A complete Swiss dossier with references and diplomas, cleanly named, is already a differentiator, because part of the flood consists of one-click applications with no attachments.

4. A contact outside the form. Marina Faria, senior recruiter at JobCloud, says she prefers a short personal first contact by phone to the motivation letter, because it gives a more authentic picture. A call with a genuine question before sending, a message to the person named in the ad, a follow-up after ten days: anything that lifts you from the list into a conversation bypasses the flood instead of swimming in it.

5. The route beside the ad. A published ad draws 300 applications; a well-researched spontaneous application draws one. When vacancies registered with the RAV have risen 13.4 percent year on year while ads on the platforms shrink, that also tells you: not every job ends up on jobs.ch.

Where preparAItor Fits

AI in an application is not a problem — generic AI is. preparAItor is built for the Swiss application, not for the global average:

• Match Check before writing: which requirements you meet, which terms are missing from the CV — so you spend your time on the jobs where you sit at the front of the stack
• Company research inside the generation: current information about the company flows into the letter and the interview preparation — the concrete details that lift the letter out of the pattern
• Swiss spelling and tone: no German ß, Swiss salutation and closing, selectable tone — the formal features by which 55 percent of companies want to recognise AI text simply aren't there
• Editable before download: the generated text is the draft; your sentence that only you could write goes in before anything leaves the platform

Sources

• JobCloud – The Swiss labour market in Q1 2026
• JobCloud – Q2 update: the Swiss labour market between war shock, recovery and structural change
• AMOSA – Artificial intelligence in the job search (2026 study)
• Netzwoche – AI opens more doors to interviews for job seekers
• Michael Page – AI reshaping Swiss recruitment: Talent Trends 2026
• HR Today – Recruiting underrates experience and lags on AI and salary disclosure (BEST RECRUITERS 2025/26)
• Robert Walters – AI makes the number of applications explode
• Fedlex – Federal Act on Data Protection (FADP), Art. 21 and 25
• arbeitsrecht-aktuell.ch – AI in the application process: may the employer have CVs pre-sorted automatically?
• EUR-Lex – Regulation (EU) 2026/1744 (Digital Omnibus)
• SECO – The labour market situation, July 2026
• SBB – FAQ for applicants
• jobs.ch Job Coach – The motivation letter: a thing of the past?