Unknown Actions in Power Automate day 5: Deepl (Preview) actions

Exploring Deepl (Preview) in Power Automate  

Power Automate’s Deepl (Preview) connector leverages the powerful DeepL translation engine—praised for its accuracy and contextual understanding. In its Preview stage, this connector enables two main use-cases: translating inline text and entire documents. This flexibility is key for automating multilingual workflows, whether you're handling user-generated content, internal communications, or dynamic document translation.  

Think of it as your bridge across language barriers: whether translating email bodies, SharePoint pages, or PDF attachments. In this article, we'll uncover all the built-in actions available and how to put them to work, one feature at a time.  

Translate Text Action  

The Translate text action is the core functionality. It takes a text string, identifies or lets you specify the source language, selects the target language, and returns a translated text.  

Inputs include:  

• Text : Raw content or dynamic value  
• Source language : Optional (auto-detect is default)  
• Target language : Mandatory (choose from DeepL’s supported languages like DE, FR, ES, etc.)  

Outputs deliver:  

• Translated text : Ready to use in emails, notifications, or systems  

Use cases include:  

• Translating user onboarding messages on the fly  
• Auto-responding in the recipient’s language  
• Converting chatbot inputs for global customer support  

Example flow: trigger on new email → extract body → translate to English (if original is non-English) → forward translated summary to a central service.  

Translate Document Action  

Going beyond snippets, Translate document supports full-file transformation. Currently compatible formats include Word, PowerPoint, and PDF.  

Inputs :  

• Document content : Typically a base64–encoded file from SharePoint, OneDrive, etc.  
• Source language : Optional (auto-detect supported)  
• Target language : Required  

Outputs :  

• Translated document content : In original file format, base64 encoded  

Use cases:  

• Auto-generating multilingual documentation or marketing materials  
• Translating uploaded documents for global review  
• Providing localized contracts or templates based on user input  

Example: user uploads a PowerPoint to a OneDrive folder → flow runs Translate document → places translated file in a language-coded subfolder.  

Glossary Management Actions  

Advanced translation scenarios often require brand-specific or domain-specific terminology. Deepl supports glossaries which map preferred terms across languages. The connector provides three glossary actions:  

• Create glossary : Build a term map between source and target languages using simple CSV-like format "sourceTerm=targetTerm" per line.  

• Get glossary : Retrieve glossary metadata—name, ID, source/target languages, created/modification timestamps.  

• Delete glossary : Remove an existing glossary to clean up or refresh terminology sets.  

These actions become critical when translating technical content: keeping product names, domain jargon, or legal phrasing consistent. Glossaries integrate seamlessly: specify the glossary ID on translate actions to ensure accurate and standardized translations.  

Get Usage and Account Info Action  

For organizational transparency and cost management, this action exposes your DeepL API usage:  

Outputs typically include:  

• Character count : Tracked vs. plan quota  
• Billing info : Currency, total spend  
• Limit alerts : Percent or absolute thresholds  

A practical use case: set up an alert flow—if usage exceeds 80% of monthly quota, email the admin or pause translation processes until reviewed.  

Detect Language Action  

Ensure correct translation interpretation by using the Detect language action. Feed it a text snippet:  

Input :  

• Text : Up to a max length (e.g. a sentence or paragraph)  

Output :  

• Language code : ISO code with confidence score  

Why useful: helps you decide whether to translate at all, or select human review if confidence is low. Example: build smart translators that skip English text while translating everything else.  

Error and Rate-Limit Handling  

Though not formal actions, it’s essential to understand how errors flow through and how to gracefully manage them in Power Automate:  

• Many actions emit standard HTTP-like error codes. You can wrap them in Configure run after logic or use Scope with On failure branches to catch issues.  
• Rate limits or quote exhaustion will show up via specific response headers or codes—your usage action or direct errors help spot and route these.  
• Standardize retry strategy: exponential backoff with delays, but avoid automatic retry on unsupported content errors.  

Sample Flow: Translating New SharePoint Documents  

Imagine a scenario where your global marketing team drops a new file into SharePoint, and you need automatic translation into 3 languages: French, Spanish, Japanese.  

Flow sketch : - Trigger : When a file is created in SharePoint folder - Condition : If the file type is .docx/.pptx/.pdf - Translate document actions: one per target language, all fed from the same file - Post-processing : Save each translated output back into language-specific folder - Glossary usage : Specify previously created marketing glossary in each action - Notification : Final step sends confirmation email with translated files  

This flow exemplifies modularity—you can easily add more languages or alternate storage steps as needed.  

Integrating with Other Systems  

Power Automate’s real strength is orchestration. Once translated, the text or document can feed:  

• Email : using built-in Outlook or Gmail connectors  
• Chat apps : Slack or Teams connectors to broadcast translation-ready updates  
• CRM/ERP : create or update multilingual records with translation payload  
• Databases : store translation metadata and stats in a logging table  
• AI systems : pass content into AI sentiment or voice systems post-translation  

Plugging translate actions into larger workflows transforms translation from a solo task into a scalable, fully integrated enterprise service.  

Best Practices & Tips  

Embrace these to get the most from the preview connector:  

• Batch text translation : Concatenate short phrases together if each API call has overhead • Glossary refinement : update term mappings before large batches to ensure consistency • Language filtering : always auto-detect or detect keyed-in language before translating • Quota awareness : run a weekly usage log to prevent surprise billing • File type limits : watch out for file size; split PDFs if needed • Failure handling : build retry scopes but escalate persistent failure to human monitors  

Future of Deepl Connector  

Because it’s in Preview, keep an eye out for enhancements:  

• Additional actions like listing available glossaries or language pairs  
• Expanded document formats support (like HTML or plain text files)  
• Native integration with glossary actions in UI  
• Throttling policies exposed as flow configuration options  
• Bulk file translation action to run multiple formats in one step  

These will make translation workflows even more robust and easier to manage.  

Summary of Deepl Preview Actions  

Below is a consolidated list of all available actions in Deepl (Preview):  

• Translate text  
• Translate document  
• Detect language  
• Create glossary  
• Get glossary  
• Delete glossary  
• Get usage/account info  

Each action can be used standalone or combined across flows. Paired with Power Automate’s native logic, connectors, and error handling, these actions let you architect profession-grade multilingual workflows—from simple auto-translations to enterprise content pipelines.  

Wrapping Up  

At Day 5 in our series, we've turned the spotlight on Deepl (Preview) —a surprisingly rich addition to Power Automate’s translation toolbox. Whether you’re translating chat messages, bulk documents, or enforcing branding terms, this connector handles the heavy lifting, while you focus on flow design, integrations, and user experience.  

Play around with each action. Try translating text, then expand to documents. Toss in a glossary and see your translated materials come out aligned with branding. Finally, add usage monitoring so you keep translation cost-aware.  

Tomorrow we’ll unpack another “unknown action” and how to leverage it in your automation stack. Stay tuned—and keep exploring!   

Source: Unknown Actions in Power Automate day 5: Deepl (Preview) actions