Introduction
Elena Grau received the rejection on a Tuesday morning. The agency needed less translation, more “post-editing.” The word sounded harmless, almost clean. For Elena it meant fewer assignments and more responsibility for texts that a machine had made faster, but not wiser.
She lives in Berlin and now advises organizations that use AI translation without pushing responsibility out of the process. Her specialty is the small difference between correct and suitable.
Story of the Path into AI
At first Elena was angry. Machines translated campaign texts, health information and internal emails at a speed no person could match. But the mistakes were not random. A polite hint became too harsh in one language, an inclusive term suddenly sounded bureaucratic, a metaphor travelled on as nonsense.
Instead of arguing only against the tools, Elena collected their weaknesses. She learned terminology management, quality metrics and how to work with language models. Her first project was a review workflow for a nonprofit that published information in several languages. AI was allowed to provide drafts, but every delicate place was marked: tone, register, cultural reference, possible misunderstanding. Once the system translated a neutral health notice as if it were giving an order. Elena showed the team why grammar is not enough.
Current Work
Today Elena builds localization processes for associations, public offices and small companies. For a flyer about counselling sessions, she let the machine suggest three versions. Then a native speaker checked the tone, a professional checked the content and Elena checked the consistency of terms. This takes longer than one click, but much less time than starting over after a failed publication.
Elena does not see her work as saving an old job profile. Translation is shifting: less first draft, more context review, terminology and risk assessment. Hiding that makes translators smaller. Designing it well gives their knowledge a new place.
Personal Advice
“Your expertise does not disappear; it sits somewhere else in the process,” Elena says. Her advice: do not fight every tool, but ask every automatic text who it sounds as if it was written for.
Key Facts
Age and place: 39, Berlin.
Background: translation, unemployment, professional renegotiation.
Entry into AI: review workflow for multilingual information materials.
Focus today: AI localization with human quality control.
Typical tools: machine translation, terminology databases, quality metrics.
Werkstattnotiz
Elena keeps a list of beautiful wrong sentences. One of them was grammatically flawless and still missed the tone of its audience. She reads the list aloud before new projects. It reminds her that localization is not finished at the dictionary; the decisive point is where a text is supposed to reach someone.