Today
At a customer's plant in Poland, Spain or Mexico, troubleshooting often fails on language, not technology. The shift leader explains the fault in his language, the technician understands half, and a translation app turns technical terms into nonsense because it does not know your part names and fault patterns. What was agreed then lives only in one of their heads.
An example
A toolmaker from Norvelta stands in an injection molding plant in Poland at an 8-cavity mold for 2K closures that is flashing at two cavities. The shift leader explains in Polish that the flash appeared only after the mold moved to another machine; demi interprets and uses the terms for parting line, hot runner nozzle and ejector from the Polish operating manual Norvelta had translated two years ago. Twenty minutes later it is clear that the clamping force on the new machine is set too low. The report with cause, new setting and photos of the cavities is then ready in German and Polish.
Your data, wherever it lives
Your technical language lives today in translated operating manuals and spare part lists as PDFs in the DMS, in training material on the wiki and in old service reports; your customers' machines sit in the maintenance system. demi reads, understands and links all of it into a glossary per language and machine, with your rights. AI interprets on site sentence by sentence with your terms and writes the report in both languages; technician and customer decide what is agreed.
Where the data lives today
- Phone & voice
- Photos, video & audio
- PDFs
- DMS & archive
- Wiki & intranet
- Tickets
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
Data sources in this case: field service app, DMS, wiki and maintenance system. demi replaces nothing; from your released translations of operating manuals, spare part lists and training material it builds a glossary for each language. It adds the customer's machine, configuration and service history so that terms and parts match the right machine.
- 2 · PermitPermission Layer
Interpreting and transcription run only if both people agree. The customer gets the report in their language; the technician's internal notes stay internal. The report reaches the ticket only after the technician has checked it, and every commitment is recorded with time and speaker.
- 3 · ActIntelligence Layer
The agent interprets sentence by sentence in both directions and uses the terms from your glossary for parts, faults and units. Numbers, dimensions and part numbers are passed on unchanged and shown on screen to read along. If a term is ambiguous, it asks instead of guessing.
How it runs
- 1
The technician starts demi on their phone and picks the customer's language; both agree to the transcript.
- 2
Each speaks their own language; demi interprets sentence by sentence, aloud and as text on screen.
- 3
Part names, faults and units come from your glossary; numbers and part numbers stay unchanged.
- 4
Photos and readings either of them shows are attached at the right point in the conversation.
- 5
At the end the report exists in both languages; the technician checks it and the customer gets their version.
What changes
Clear communication when troubleshooting, even without a shared language
Agreements recorded in both languages
Fewer queries and repeat visits abroad
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
+108 hfreed per year
Assumptions and calculation
demi interprets on site with your glossary and writes the report in both languages; you check it.
60 min → 42 min per case
How we calculate
30 foreign-language service visits × 60 min × 30% × 1.0 (manual) × 12 months = 6,480 min = 108 h
Value per year, before demi costs
Freed per year+108 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
+$7,020
gained per year
108 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 packageQuestions about this use case
Which languages can demi interpret?
The common languages of your customers and suppliers, such as English, Polish, Czech, Spanish, Italian, Turkish or Chinese. It works best in languages for which released translations of your documents already exist, because the glossary is built from them. For a new language, you add the key terms once.
How does demi know our technical terms in the other language?
From your released translations: operating manuals, spare part lists, training material and earlier reports. demi builds a glossary for each language from them, which you can check and extend. If a term is missing, demi shows that and uses the general translation.
Do we need special equipment for this?
No. A company phone is enough; at loud machines a headset helps. The text also appears on screen, so numbers and dimensions can be read along. When several people speak, demi assigns each sentence to its speaker.
What happens to the conversation afterwards?
The checked report is kept in both languages, with photos and readings attached to the ticket. The audio is deleted once the report is released. demi proposes new terms from the conversation for the glossary; they are added only after review.
