DeepL can perform automated steps in your workflows (see Automated BMS steps). BeLazy sends DeepL the XLIFF it exports from your TMS.


Full-text translation

BeLazy has DeepL translate the full text of the document, not segment by segment. DeepL can therefore take the surrounding text into account, instead of translating each segment on its own, and the segments are filled back into the XLIFF afterwards.


Adding DeepL

Open Translation Providers in the profile menu and click Add to add a DeepL provider. Enter a Name for it and your DeepL API key. The card then shows "API key configured". On the card, the pencil icon edits the provider, the bin icon deletes it, and the settings icon opens its setup, which has three steps: Credentials (your API key), Language mapping and Manage presets.


Language mapping

This maps the languages of BeLazy to the languages of DeepL, so that BeLazy knows which DeepL language to use for each of your languages.


Presets

A preset is a saved set of DeepL options that you choose in the workflow mapping. Under Manage presets, each preset has a Name, a Formality (Default, Less formal or More formal) and an optional Glossary ID, the ID of the DeepL glossary to use. Add as many presets as you need, for example one per customer. Use the pencil and bin icons to edit or delete a preset. In the workflow mapping, the presets appear under the name of your provider.


How BeLazy saves AI costs

BeLazy keeps your DeepL usage down in three ways:

  • No minimum character charge. BeLazy sends DeepL the content of the XLIFF, not your source files, so DeepL's minimum of 50,000 characters does not apply.
  • Nothing is paid for twice. Non-translatable parts and locked parts are filtered out before the text goes to DeepL, so they are not charged a second time.
  • It fits translation memory preparation. This makes DeepL work well with translation memory preparation: the parts your translation memory already covers do not need to be paid for again.


Tags

BeLazy has its own mechanism to resolve tags, whichever TMS the XLIFF comes from, because the DeepL API does not handle tags well.