Skip to content
Knowledge & documentationPumps & ValvesElectronics & SensorsMachinery & Plant EngineeringConveying & Intralogistics+2

Turn old translations into translation memory

Translations you paid for become a memory: demi aligns them sentence by sentence, and only what changed gets translated again.

Today

Manuals, data sheets and quotes were translated into many languages over the years, by changing agencies, and sit in folders apart from their originals. There is no translation memory, or the agency keeps it. With every new edition you pay again for sentences translated long ago, and the same terms come back slightly different each time.

An example

Valbrecht has had the instructions for use of the VZ 248 infusion pump translated into 14 languages by three agencies since 2016, most of it only as PDF. demi aligns 41,000 sentence pairs and finds three different Italian terms for "occlusion alarm". Only 620 sentences are new or changed for the new edition, and demi reports that the Finnish version is still at its 2021 revision.
Fictional company, typical workflow.

Your data, wherever it lives

Your past translations live today in folders and the DMS, as Word files, PDFs and files returned by translation agencies over many years, mostly apart from their originals. demi reads, understands and links them sentence by sentence, with your rights. With every new edition the AI knows at once what is already translated and what is new, and one person per language approves.

Where the data lives today

  • Folders & drives
  • Office files
  • PDFs
  • DMS & archive
Data LayerOne view of everything, with your rights

How demi solves it

Three layers in this order: connect the data first, then the rights, then the AI.

  1. 1 · ConnectData Layer

    Data sources in this case: the file share, DMS and PDFs, plus the files returned from past translation jobs. demi replaces nothing, finds every original's language versions and aligns source and target text sentence by sentence. Each sentence pair knows its document, revision, language and approval and can be handed on in the standard TMX format.

  2. 2 · PermitPermission Layer

    Technical documentation maintains the memory. Sentence pairs from confidential documents such as quotes are used only by people allowed to see the original. Every approval and every change to the memory is recorded.

  3. 3 · ActIntelligence Layer

    The agent matches sentences by numbers, part numbers, structure and meaning and flags uncertain pairs for review. When the original changes, it shows which sentences are affected in which language version and translates only those, with glossary and memory. A fixed rule set then checks numbers, units and part numbers.

How it runs

  1. 1

    Connect folders, the DMS and the files of old translation jobs

  2. 2

    Find each original's language versions and align them sentence by sentence

  3. 3

    Review and approve uncertain pairs and conflicting terms

  4. 4

    For a new edition, translate only new and changed sentences

  5. 5

    A reviewer per language approves, and the memory grows with each job

What changes

  • Translations you paid for are reused

  • Same sentences and terms in every language version

  • You see which language version lags behind the original

What does it bring you?

Check it with your own numbers

Every starting value is an assumption you can change. The calculation sits right below, step by step.

Assumption, editable

Your numbers

Result

+216 hfreed per year

Assumptions and calculation

demi aligns old translations and sends only new and changed sentences to translation; you approve.

3 h → 2.1 h per case

How does it run today?

What is already digital saves less: the factor lowers the share demi brings.

How we calculate

20 language versions × 3 h × 30% × 1.0 (manual) × 12 months = 216 h

Value per year, before demi costs

Freed per year+216 h

Assumption: labor cost of $48.62 per hour in US manufacturing in June 2026 (BLS), plus workplace, IT and overhead.

= Value per year, before demi costs

+$14,040

gained per year

216 h ÷ 1,760 h per role = 0.1 full-time roles

An estimate, not a promise. The starting values are assumptions for an industrial company with 150 to 500 people.

Combine with more cases and get the package

Questions about this use case

Does this work when old translations exist only as PDFs?

Yes. demi reads PDFs, scanned ones too, and text files, and matches each original with its language version by title, part number, revision and structure. Where the match is uncertain, for instance because a version was shortened, the agent flags the passage instead of guessing.

Can our translation agency use the memory?

Yes. demi exports it in the standard TMX format that common translation tools read. The agency then sees for every sentence whether it was translated before, and matches from the memory are usually charged at a lower rate. Whether you translate in-house or with the agency is up to you.

How does demi know which sentences changed in the original?

The agent compares the new version sentence by sentence with the last approved revision. Unchanged sentences come from the memory; new and changed ones go to translation. For each language version you see its revision and which sentences it is missing.

What happens to conflicting old translations?

The agent puts the variants side by side, with document and year. Technical documentation decides which one applies, and the others are blocked in the memory. That way an old mistake does not travel into the next edition.

See this case with your data

In the demo we show you the workflow on your own systems.

Your configurator

No use cases yet

Your configurator is still empty

Add a use case from any card or case page and see what it brings per year, with the calculation. Free, no sign-up, no commitment: try it out and change any number.

Browse use cases