A Truly Local ATS, in 75 Languages for hiring teams in Saatli, Saatly DistrictAzerbaijan
A practical guide for recruiters working in Saatli. View without a localization
A Truly Local ATS, in 75 Languages
75 native languages alongside English, live today — and a Candidate Match Score that reads every one of them.
Most hiring software is built in English and translated afterwards, if at all. The interface may appear in another language while everything underneath — the matching, the scoring, the job feed — still assumes English input. That gap is invisible until a German employer pastes a German CV against a German job description and gets a screening result that quietly understood neither.
Expertini approaches this as two separate problems, because they are two separate problems. The first is localisation: rendering the product in the language of the person using it. The second is semantic understanding across languages: reading a document in one language against a requirement written in another. Solving the first does nothing for the second, and a great deal of recruitment software solves only the first.
Expertini ATS runs in 75 native languages alongside English today, and the Candidate Match Score — our own scoring layer, not a bought-in service — reads all of them. That combination is still uncommon in this category: plenty of platforms will show you a translated menu, far fewer will read a Portuguese CV against a Dutch job description and tell you, with citations, why it does or does not fit.
On this page
01What 'localised' means here, concretely
Each language runs as its own deployment on its own host, so a visitor is never served a partially translated page assembled from two sources. The interface, the transactional e-mails, and the marketing pages are each built per language from the English source, and every page is verified before it ships: the templating structure, the embedded JavaScript and CSS, and the placeholder positions inside every sentence must survive the translation exactly. A translation that would break a page is refused and the English is kept instead.
Beyond the interface, the data layer is localised too — job categories, employment types, experience levels, industries and country names. What is deliberately *not* translated is the identifier underneath each of those. `full-time` stays `full-time` and `computer-and-mathematical` stays `computer-and-mathematical` in all 75 languages, because those identifiers are how a job published in the ATS is matched onto the country job sites. Translating the label is localisation; translating the identifier would silently disconnect the job from the network it is meant to reach.
The same distinction applies to structured data. Experience level is shown to a recruiter in their language but transmitted to Google for Jobs in the canonical English value, because that is what the schema expects. A localised interface should never change what machines are told.
02Understanding a language is not the same as displaying it
The harder half is semantic. A candidate writes their CV in their own language; an employer writes the role in theirs; they are frequently not the same language. Keyword matching fails here completely — `Teamleitung` and `team leadership` share no characters — and this is precisely the case where reading for meaning rather than for strings earns its place.
In Expertini's screening path, the AI layer reads the CV against the job description and extracts evidence for each competency the description actually asks for. That extraction is explicitly cross-lingual: a German CV read against an English job description yields English evidence citing what the German document said. The recruiter reads the result in their own language; the candidate is never asked to write in someone else's.
What does *not* change across languages is the score. The Candidate Match Score is produced by a fixed published formula from the extracted evidence, not generated by the AI — so language affects what evidence is found, never how that evidence is arithmetically weighed. A candidate is not scored differently for having written in Portuguese.
03Which language the software answers in
Generative features — drafting a job description, writing candidate correspondence, the support assistant — have a subtler failure mode. Every instruction the platform sends its AI layer is written in English, and a model can mistake that for an instruction about which language to *reply* in, or drift on incidental cues in the content. A job title mentioning a European city is not a request for a European language.
Expertini therefore separates the two explicitly. Platform instructions are never a language signal; only the employer's own words are, and they are marked as such when they are sent. Type a job title in German and the draft comes back in German; type it in English and it comes back in English, whichever city it mentions.
Screening output follows a different rule on purpose, because the reader is different. There, the language belongs to the hiring team — the people who will read the report — and the candidate's CV is treated as evidence to be understood rather than a voice to be echoed back.
04What this means for a hiring team
The practical effect is that language stops being a filter on your pipeline. A candidate in Kraków applies in Polish, a recruiter in Munich reads the evidence in German, and the score underneath is produced by the same published formula either way. You are not choosing between hiring locally and hiring across a border, and you are not paying a translation agency to find out whether a CV was worth reading.
Every language is a complete deployment, not a language pack bolted onto an English product. The same features, the same integrations, the same pricing — there is no international tier and no per-market surcharge. Teams that already hire in several countries usually run several language hosts side by side against one account, because the jobs, candidates and billing are shared.
Two things stay in English on purpose, and both protect your distribution. A small number of strings are kept in English where translating them would break a placeholder. And city names are never translated: they are the identifiers behind hundreds of thousands of location pages and the match onto the country job sites, so localising them would quietly disconnect your jobs from the network they are meant to reach. If you spot wording that could be better in your language, contact support and we will correct it.
Frequently asked questions
Do the language sites cost extra?⌄
Does a candidate have to write their CV in the employer's language?⌄
Does the score change depending on the language?⌄
How does advanced semantic matching help across languages?⌄
Does publishing in one language limit where the job reaches?⌄
We hire in several countries — do we need separate accounts?⌄
What if some wording could be better in our language?⌄
At a glance
- Our own Candidate Match Score — now reading in your language, not just English
- Advanced semantic matching across languages, still uncommon in this category
- A Polish CV read against an English job description, evidence cited either way
- 75 languages, each a full deployment — no international tier, no surcharge
- One account and one candidate database across every market you hire in
- Distribution unaffected by the language you publish in
See a truly local ats, in 75 languages on your own hiring.
Bring a real job description to a 30-minute demo — free trial included.
Book a demo