Official-level AI translation.
Accurate. Consistent. Verifiable.
By linguists, for professionals.
What you receive
Official-level means held to the standard of professional human translation.
Your document, translated: translator-grade work, ready for your final check.
Every passage checked against the source: the findings, the verdict, and what to do next.
The key terms of your document, researched before translation and locked for consistency.
Checked against the source, passage by passage.
How it works
Send your document and requirements. You receive a delivery date.
A linguistically driven seven-step pipeline mirrors what a professional translator does, with verification gates at every step. All seven run at AI speed.
The translation comes back with its QA report and terminology report. You check, decide and sign off: with a second pair of eyes, or on the evidence alone.
Quality & responsibility
There are many AI translation platforms. Translyta is built for the one thing they do not deliver: quality a professional can stand behind. The reason is the pipeline, designed by linguists on the principles of the craft: it works like a professional translator, not like a statistical machine.
And we are precise about what Translyta is not: no human has checked the translation before it reaches you. Professional translation has always worked with two pairs of eyes: one produces, one judges. Translyta is the first pair: it translates, revises and documents. The second pair is human and yours: you give the final sign-off.
Who it’s for
Most business documents never need a certified translation: reports, policies, manuals, websites. They still need to be right, and to keep one term for one concept, first page to last.
When the wording carries legal effect, a plausible translation is not enough. Your in-house revision starts from researched terminology and a documented check against the source.
Scale production without scaling cost. Your translators and revisers start from translator-grade work instead of raw machine output, with the terminology already researched and locked.
Skip the first draft. Start where your judgment matters, with the terminology researched and the QA findings already in hand, so your scarce hours go to the calls only you can make.
Every delivery arrives ready for revision: the translation itself, plus its researched terminology and a report of the checks and findings. Your reviser, a subject expert, or you: whoever signs off does so on evidence, not on faith.
If you know MTPE: this is what post-editing should have been. Not repairing raw machine output, but final revision on translator-grade work.
What makes us different
Machine translation, translation memory, general-purpose AI: they all treat translation as a technical problem. Translyta treats it as a linguistic one. Every step that touches the language applies linguistic expertise, not just statistical optimization.
That expertise rests on four principles, ranked so a conflict has an answer: accuracy, completeness, clarity, consistency. Where accuracy and fluency pull apart, accuracy wins.
The discipline behind those calls is written out in our public working reference: 158 entries on translation techniques, language structures, terminology, phraseology and conventions, cross-linked and sourced.
About
Translyta grew out of Dragoman Translations, an official and legal translation practice running since 2001. The pipeline is that practice rebuilt for AI: the same steps, in the same order, on every document.
Translyta was selected for Tech Istanbul’s 2026 pre-incubation program and is being set up as a company of its own. Until that company is in place, your contract and your invoice are with Dragoman.
Comparative linguist (Leiden University), LangOps Pioneer and sworn translator, with published research on Turkic morphology. Lecturer and examiner at SIGV, the Dutch institute that qualified court interpreters and legal translators. Supported by a team of three in the Netherlands.
Maarten designed the pipeline and wrote its core: the principles, the rules and the linguistic profiles. He built a database-driven system for terminology research and maintenance.
He argued the case in a keynote at the 2026 Global Language Services Conference: translation is not a technical problem but a linguistic one. He tested the pipeline against the same exam papers he once graded, and kept rebuilding until it passed them all.
What the pipeline has done
Contact
We are running our first proofs of concept with a small number of companies. Want to join? You will see exactly what a Translyta delivery looks like: translation, QA report and terminology report included.
Into Dutch: from German, Turkish, Azerbaijani and Uzbek.
Into Turkish: from Dutch and English.
Into English: from Turkish.
Ask about other language pairs.
Translyta is a business-to-business service.