You decide the exact word your company's glossary terms appear with in each language. The app imposes it within the translation and then programmatically verifies its presence.
A translator who chooses well each time chooses differently each time. Here, the word is decided only once and repeated always.
Fixed terminology is the «translate it ALWAYS this way» mode of the glossary. You add a glossary term in your source language and write, language by language, the word you want it to appear as: «checkout» → «finalizar compra» in Spanish, «Kasse» in German, «panier» in French. That pair is tied to the glossary term, not to a specific text, so it applies equally to today's email and to the product sheet three months from now.
It is not a polite instruction to the translator. Before your text goes to the model, each occurrence of the glossary term is replaced by an internal marker; when the translation returns, that marker is replaced by the word you fixed. The model never gets to decide on that glossary term, and the rule also travels in the instructions so that the sentence’s environment (articles, prepositions, gender) agrees well.
Afterwards, it is verified. The final text is checked for how many times the word you imposed appears; if it's not there, there are up to two targeted retries, and if it's still not achieved, the report declares it instead of disguising it. The initial capital letter adjusts automatically to the context: if the glossary term falls at the beginning of the sentence, it appears with a capital letter, and in the middle, it appears as you wrote it.
And you don't have to fill in all six languages manually. The «Propose translations» button asks the AI for the word for each language at once, using your own text as context —the «checkout» of a store is not that of a hotel—. The suggestions arrive in amber, marked as UNCONFIRMED: they are not your terminology until you tap them or click «Accept all».
Use it free without an account. If you want to save your result, you can sign up afterwards.
Save / Sign upOne word per language, chosen by you
each glossary term has an input field for each target language. You write the equivalent your company uses, and that's what appears, not the synonym the translator might prefer today.
It is imposed beforehand, not pleaded for afterwards
the glossary term travels masked, and the fixed word is placed upon restitution. The model cannot ignore a rule it doesn't see.
It's checked with code, term by term
the imposed word is counted in the final text. If it's missing, there are up to two targeted retries; if it's not achieved, the report indicates it with your term and the specific language.
Capitalization adjusts to context
the imposed term appears with an initial capital letter when starting a sentence and in lowercase when within it, so your text doesn't look like an error.
Assisted suggestions for all six languages at once
a button fills all fields. The AI works with the context of your own text, not with the isolated term.
Suggestions are visible and validated
they arrive in amber and italics, marked as unconfirmed. You edit them one by one or confirm them all; until you do, they don't count as your terminology.
You can leave languages without a rule
an empty field means «the translator decides here». You only fix the languages and terms that truly matter to you.
The same word on the web, in the app, and in emails
the glossary is saved with a free account —up to 20 glossaries of 400 terms, importable and exportable in CSV— and is reused in each translation, not rewritten every time.
Mark a term, write its fixed translation, and check the count in the report. The first three translations don’t require an account.
In the features that use AI, responses are generated by artificial intelligence and are not a substitute for professional advice.