What an AI CV builder can and cannot do for a Swiss CV

A Swiss CV has a shape. Personal details at the top, including nationality or permit and date of birth. A photo, still expected by most employers. Experience in reverse order with month and year. Education, then languages with a level, then skills. Two pages for most people, sometimes three for a long career. None of that is law, it is convention, and conventions are exactly what a language model has read thousands of times. Give it a US resume or a Europass and ask for the Swiss version and you get the right order back in seconds.

The second thing it does well is vocabulary. Adverts in the same field describe the same work in different words, and the first reader of your CV is often a system that matches words against the advert. If the advert says "Projektleitung" and your CV says "project coordination", a keyword match can miss it. Pasting the advert next to your CV and asking the tool to use the advert's terms where they are true is a legitimate, quick improvement.

Where it is true is the condition. The system that sorts and the person who decides are different readers. The person who opens the CV afterwards looks at where you worked, what you delivered there and whether the skills fit. None of that improves by adding more terms, and a CV stuffed with the advert's vocabulary reads to that person as copied. Use the tool to make the right facts findable, not to add new ones.

Swiss adverts come in several languages, and the advert's language is usually the language the dossier should be in. Counted where the language of the advert could be identified, 56% are written in German, 26% in English, 14% in French and 2% in Italian. Anyone applying across the language border or from abroad needs at least two versions of the same CV, and producing a clean second version is the job AI tools are genuinely good at. The output is fluent and consistent, which a hurried manual translation often is not.

Three things still need a human. Swiss qualifications should keep their Swiss names: an EFZ, a Matura or an HF diploma translated into a foreign equivalent loses the meaning an employer reads in it. Put the original name and add a short explanation in brackets if the reader might not know it. Job titles are the second: a model will often turn a precise title into a generic one, and the title is one of the first things a recruiter reads. Company names, finally, stay exactly as they are.

And give the tool the advert as well as the CV. A translation that uses the advert's terms for your work reads as written for the job, while a neutral translation reads as translated.

This is the most common misunderstanding we see among people applying from abroad, and AI makes it easier to act on. An English advert from a Zurich employer looks like an open door, the tool produces a polished English CV in a minute, and the application goes out. But across live Swiss adverts written in English, 23% also name German as a requirement. The advert is in English because the team writes in English; the office, the customers or the canton still work in German.

The same holds in the other direction: a French-language advert from Geneva or Lausanne asks for German in about one case in five as well. So before you ask any tool for a translation, read the language line in the advert. If it names a language you do not have at the level it asks for, a better CV will not change the outcome, and your time is better spent on the adverts whose languages you actually have.

If you do have the second language, show it on the CV where it counts: in the languages section with a level, and ideally in one line of experience where you used it at work.

Ask a model to make your CV more competitive and it will quietly upgrade what it can: "good German" becomes "fluent German", and an honest B1 drifts towards C1. It does this because it has no idea what your German is like, and because stronger words are what "improve this" usually means. Check the languages section line by line after any AI pass.

The reason it matters in Switzerland is what the adverts demand. When a Swiss advert names German, 80% of the time it asks for professional or native level; for English the figure is 86%, for French 79%. Those are working levels, and an employer asking for one has a reason to check it, usually by switching language a few minutes into the interview. A level that does not hold up for ten minutes of conversation ends the process, and it makes the rest of the CV look doubtful too.

So write the level with a CEFR grade and a piece of evidence: a certificate, a year of work in the language, a degree taught in it. "German B2, Goethe certificate 2024" is more useful to a recruiter than "fluent", and no tool can make it up for you.

Look at what a Swiss advert states. 85% say what kind of contract it is and two thirds name a language. Not one of the Swiss adverts we read gives a salary figure. That is a measured absence, and it has a consequence people rarely hear: no tool, ours included, can benchmark a Swiss salary from advert data, because the data does not exist. A chatbot that answers "what should a marketing manager in Basel earn" is repeating whatever it read online, usually without a date or a source.

Salary matters for the CV and the letter because many Swiss application forms ask for a Lohnvorstellung, and the question comes up early in the first interview. The number you give should come from the federal wage survey, which is the only broad source of actual Swiss pay. The SGB Lohnrechner is built on it: put in canton, sector, age and function and it returns a median. Use that as a floor and give a gross annual band.

Where AI is useful here is the wording around the number: a clean sentence that states the band, says whether it assumes twelve or thirteen monthly salaries, and names the Pensum it is based on.

A Swiss application is more than a CV. It usually includes the Arbeitszeugnisse, the written references every Swiss employer issues at the end of a job, and they describe your tasks and how well you did them. So an achievement on the CV that the reference does not mention, or a responsibility that sits two levels above the job title in the reference, is visible to anyone who reads both. AI-polished CVs fail here more than anywhere else, because the tool writes confident lines about impact that the paperwork does not support.

The other giveaway is sameness. Ask a model for achievement bullets and you get "drove cross-functional initiatives" and "delivered measurable results" with no measure in them. A recruiter who reads a stack of CVs a day has seen that sentence many times this month. Concrete lines with a number, a system or a customer in them read as yours: "moved 40 clients from the old billing system without an outage" is something only you can write, and the tool can only tidy it.

So the workflow that works is the reverse of the common one. Write down the facts first: employers, dates, what you did, what changed because of it, with numbers where you have them. Then let the tool do the Swiss format, the advert's vocabulary and the translation. Then check every line against your references and your real language level before it goes out. The tool saves you an hour. It does not stand behind the document in the interview. You do.