No agency could take this job. So we built the translator.

The hard part was never one sentence

Any good translator can handle a paragraph of EU regulation. Nobody can handle tens of thousands of them, all in the same specialised language, with every term reading the same way from the first document to the last.

Who needed it, and why

The Hungarian government set up MFOI to get more Hungarians into directly managed EU programmes and to help applicants along the way. That meant tens of thousands of official EU grant and funding documents had to exist in Hungarian.

Legal, financial and regulatory text, dense with specialist terms. Traditional translation agencies had neither the capacity for that volume nor the domain knowledge it called for. The usual route was closed.

What we built

One system, trained for one job:

• A model that speaks the field. We tuned the language model to EU legal, financial and regulatory text.
• A pipeline built for volume. Our own translation pipeline for large-scale document processing.
• One concept, one term. Terminology management, so the same concept reads the same way in every document.
• Checks that run themselves. Quality assurance is automated too.

It translates from any official EU language into Hungarian.

What it changed

• Nearly 100 times more cost-effective than a traditional translation agency.
• Minutes per translation instead of weeks.
• Tens of thousands of documents handled, with room to scale beyond 100,000.
• Specialist terms stay consistent across every document.

Nearly 100 times cheaper than a translation agency, and it handles volumes no human team could match.

If your problem looks like this one

• AI earns its keep on volume plus specialism. Either one alone, a person can handle. Together they ruled out the traditional route.
• Specialist text needs a specialist model. Without tuning to the field’s language, the legal nuances get lost.
• At this scale, consistency is the quality. Ten thousand documents with drifting terminology are just noise.