Today
Technical documents have to go into many languages fast, always with the same terms. Generic translation tools pick different words every time and sometimes change numbers or units. The review falls to the few people who know both languages and the product.
An example
Talwerk Antriebstechnik needs the 180-page operating manual for its G-400 gearbox series in Polish. The agent translates using 1,200 approved terms from the term base and flags 3 lines where the unit is already missing in the German source. The Polish reviewer clears these with engineering and releases the document.
Your data, wherever it lives
Today manuals, data sheets and quotes sit as text files and PDFs in the DMS, the PLM and folders on the drive, and earlier translations from agencies are buried in e-mail attachments. demi reads, understands and links them per technical term and language, with your rights. The AI translates with your approved glossary in minutes instead of hours, a fixed rule set checks numbers, units and part numbers against the original, and a reviewer per language approves.
Where the data lives today
- Office files
- PDFs
- DMS & archive
- PLM
- Folders & drives
- E-mails
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
The case needs data from the DMS, PLM, file share and e-mail. demi replaces nothing and builds a term base from documents and earlier translations, with one entry per technical term and its approved translation for each language. demi writes back only after approval, in the same structure.
- 2 · PermitPermission Layer
Technical documentation maintains the term base. A reviewer per language approves each translation, and every approval and every change to the term base is recorded.
- 3 · ActIntelligence Layer
The agent translates with enforced terminology and proposes new terms for approval. A fixed rule set then checks line by line that numbers, units and part numbers match the source and flags every deviation.
How it runs
- 1
Connect documents and existing translations
- 2
Build the term base from past translations and approve it
- 3
Translate the document with enforced terminology
- 4
Check numbers, units and part numbers against the source
- 5
The reviewer for each language releases; the result goes back after approval
What changes
Translations ready sooner
The same terms in every document
Numbers, units and part numbers checked
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.
Your numbers
Result
+576 hfreed per year
Assumptions and calculation
demi translates with your glossary and keeps numbers, units and part numbers fixed; you proofread.
2 h → 1.2 h per case
How we calculate
60 documents × 2 h × 40% × 1.0 (manual) × 12 months = 576 h
Value per year, before demi costs
Freed per year+576 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
+$37,440
gained per year
576 h ÷ 1,760 h per role = 0.3 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 packageQuestions about this use case
Can demi build a glossary from our past translations?
Yes. The agent aligns past translations with their originals and proposes a translation for every technical term. Where you have used different words so far, it shows the variants side by side. Documentation decides which one applies.
How are numbers and units checked?
After translation, a fixed rule set compares every line with the source: numbers, units, part numbers, references to standards. This check is rule-based, not left to the AI's judgment. Decimal separators are converted correctly per language, and every real deviation is flagged.
Who reviews the translation before it goes out?
One person per language whom you appoint, in-house or at your translation agency. They see the flagged lines first and release the document. Nothing goes to customers without that release.
Which documents can be translated this way?
Manuals, data sheets, quotes, labels and training material, as well as texts straight from the PLM. The result goes back into your system in the same structure. The term base applies to every document type alike.
