Already used AI?

AI translation review
find out what you actually have

You ran it through DeepL or ChatGPT and it reads well. The question is whether it is right — and machine output reads fluently even when it is not. We check it, fix it, and tell you plainly which it was.

  • Full review by a qualified translator
  • Errors fixed, not just flagged
  • A straight answer on fitness for use
  • Certification possible after review
  • Honest advice if retranslation is cheaper
Same-day answer

Get your quote

Send the original and the machine version, and we will reply with a fixed price.

Your documents

    What a machine gets wrong, quietly

    Machine translation has become genuinely good at everyday text. Its failures moved somewhere harder to see.

    • Terms of art

      Legal and technical terms have defined meanings. A machine picks the everyday reading, and the sentence still looks correct.

    • Numbers, dates and names

      Decimal conventions, date formats and proper names are exactly where fluent output hides small, expensive errors.

    • Tone and register

      The machine translates the words and misses the relationship — formal where you were warm, blunt where you were careful.

    • The silent omission

      A dropped negation, a skipped clause, a hallucinated addition. Fluency is not fidelity, and only a reader of both languages can tell.

    Our honest position on AI

    We use machine translation ourselves, where it genuinely helps, and always with a qualified translator revising the output — never as a way to quietly deliver raw machine output at human prices. So this is not a page against the tools. It is a page about knowing what you have before you rely on it.

    When you send us a machine translation, one of our translators compares it against the original line by line, fixes what is wrong, and gives you a straight answer: it was fine, it is fine now, or it was cheaper to retranslate than to repair. You get that answer before you commit — if the text is good, we will say so.

    One thing we will not do is review a machine translation of a contract, a medical report or anything a court will read and simply "sign off" on it. For those, the review is a full revision against the source — the same standard as our own work, because our name goes on the result.

    Happy to be of service,

    Gabriel Brunner, Founder

    How a review works

    1. 1

      Send both versions

      The original document and the machine translation. Tell us what the text is for — publication, a client, an authority — because that sets the bar.

    2. 2

      We look before we quote

      A translator samples the output first. If it is good, the review is cheap; if it is beyond saving, we tell you and quote a retranslation instead.

    3. 3

      Approve the quote

      A fixed price either way. Say yes and we start.

    4. 4

      Receive the reviewed text

      Corrected, consistent, and fit for its purpose — with certification on top where an authority requires it.

    Frequently asked questions

    Tap any question to see the answer!
    Why pay for this when the AI translation is free?
    Because the machine cannot tell you whether it is right, and neither can you unless you read both languages. What you are paying for is a qualified reader of both — and the difference between "reads well" and "is correct" is exactly where the expensive mistakes live.
    Is this cheaper than a normal translation?
    Usually, when the machine output is good. When it is not, repairing it costs more than translating from scratch — in that case we say so and quote the retranslation instead. You get one fixed price either way, before you commit.
    Can you certify a machine translation after reviewing it?
    After a full revision against the original, yes — at that point it has been checked to the same standard as our own work, and we can certify it in the form the receiving authority accepts. We do not certify anything we have not revised in full.
    Which tools did you have in mind?
    Any of them — DeepL, Google Translate, ChatGPT, or the translation feature built into your CMS or CAT tool. The review is the same regardless of which machine produced the text.
    Will you tell me if the machine translation was actually fine?
    Yes, plainly. It happens — modern machine translation is genuinely good at some kinds of text. That answer costs you the review, not the retranslation, and it means the next time you can use the machine with more confidence.
    What does it cost?
    It depends on the length, the language pair and the state of the machine output, which is why we look at a sample before quoting. Send both versions and you will have a fixed price — with no obligation.

    Not sure what the machine gave you?

    Send us both versions. We will tell you honestly — even when the answer is that you did not need us.