Translator
Translate texts and documents between 13 languages with automatic language detection and your terminology from the notebook.
The Translator translates texts and whole documents right in the browser – GDPR-compliant, on the same AI model that powers chat. You paste a text on the left, choose source and target language, and the translation appears on the right once it is complete. Your terminology from the notebook is applied – so “Hausmeister” always becomes “Caretaker” and never “Janitor”.
You’ll find the Translator in the sidebar under Tools → Translator or directly at /translate. At the top you choose between Text and Document.
What the Translator does
Translation between 13 languages plus automatic source-language detection
Translate text – the translation appears on the right once it is complete and can be edited there
Translate whole documents – Word, Excel, PowerPoint, PDF, Markdown and TXT, formatting is preserved
Terminology from the notebook: pages marked “Include in every answer” from the selected areas (Company, Teams, Project, Private) apply to every translation
Translating step by step
Open
Translatorin the sidebar.Above the left field, choose the source language. The default is
Auto-detect— Kasimir then identifies the language for you.Above the right field, choose the target language. It defaults to the other of the two interface languages (so with a German interface,
English).Type or paste your text into the
Sourcefield.Click
Translate. As soon as the translation is complete, it appears on the right underTranslation.If needed, click
Copyto put the result on your clipboard (the button briefly switches toCopied).
While translating, the button shows … and is disabled. The result appears in an editor where you can still adjust it — line breaks, paragraphs and formatting from the source text are preserved.
Swapping languages
The swap button (double-arrow icon to the left of Translation) swaps source and target language; the texts stay where they are. The swap button is locked while a translation is running.
Available languages
Language | As source | As target |
|---|---|---|
Auto-detect | Yes | – |
German, English, French, Spanish, Italian | Yes | Yes |
Polish, Dutch, Portuguese, Russian, Turkish | Yes | Yes |
Arabic, Chinese, Japanese | Yes | Yes |
Auto-detect is available only as the source language, not as a target.
Terminology and knowledge
There is no separate translator glossary any more. Instead, the Translator uses your notebook: pages marked “Include in every answer” flow into every translation – especially terminology rules.
Use the Knowledge field at the top right to decide which areas these pages come from: Company, Teams, Project and Private can be switched on and off individually. All are on by default; Project only counts once you pick a project in the project field (without a selection: “No project”).
How to maintain terminology: create a “Terminology” page in the notebook (or in the project’s notebook), write the terms as a table (German · English · note) and set “Include in every answer”. The link below the Knowledge field takes you there.
Translating whole documents
Switch to Document at the top.
Choose source and target language.
Drop the file into the field or click it (Word, Excel, PowerPoint, PDF, Markdown, TXT – up to 25 MB).
Kasimir translates the file section by section; progress is shown below (“Translating … 12/40 sections”).
When it is done, download it via Download translated file.
Formatting of Word, Excel and PowerPoint files is preserved. A PDF is returned as a translated Word document – the original page layout is not preserved.
Limits and notes
Each translation is limited to a maximum of 50,000 characters. Longer text is rejected — split it into sections in that case.
Source and target language must be different, otherwise translation is blocked.
If you start a lot of translations in a short time, Kasimir briefly slows you down.
Translations are not saved and don't appear in any history; use
Copyto keep the result.Translation runs on your company's first active chat model, so quality matches the language model used in chat.