Gokan's Studio

Is SharePoint’s Built‐In Translator Really Dead?

In modern SharePoint, AI features like AutoFill columns and Copilot have transformed how content is handled. AutoFill uses large language models to automatically extract or generate content from documents and populate metadata columns. For example, you can set up a column prompt such as “Extract the invoice number”, and SharePoint will fill it in automatically. The new Knowledge Agent in SharePoint even “intelligently tags and classifies files with auto-filled metadata” so that Copilot can use it when answering questions. In practice, this means AI now does tasks that once might have required a manual translation step.

Copilot in Office can translate on the fly – in Word, you can translate an entire document or selected text into another language, and in PowerPoint, you can choose Translate this Presentation from the Copilot menu and get a fully translated copy. In other words, for any metadata field or text snippet in SharePoint, you can ask Copilot to “translate X into [language]” and let the LLM do it. Your Copilot/Autofill subscription covers this approach, so for these use cases, a separate translation engine is essentially redundant.

• AutoFill Columns: Lets you define a natural language prompt and automatically pulls that info from files into columns. No manual tagging or translation needed – the AI handles it.
• Knowledge Agent: Continuously enriches content, suggesting autofill columns and tagging files with AI-generated metadata. You can even instruct it in different languages.
• Copilot Chat/Office: Built‐in translation features let you translate text within the flow of work. You don’t need an external translator for pages, lists, or column values – Copilot can do it instantly.

Because of this, for metadata, pages, and similar content tasks, a standalone translation service is essentially “dead.” The AI you already have (Copilot and Syntex with PAYG) can handle multilingual content natively. There’s no need to call an external translator to get field values into French, Spanish, Turkish, etc., or to classify a document’s topic in multiple languages – the LLM does it. In short, translation is obsolete for everyday AI/content operations in SharePoint, except for whole-document translation.

Document translation | the last stand

That said, the official Document Translation service isn’t completely gone – it just remains dedicated to whole‐document scenarios. The feature, powered by the Azure AI Translator Document Translation API and consumed through SharePoint Document Processing (Syntex), provides a service that “creates a translated copy of a document in a SharePoint document library, preserving the original format and structure.” Through OneDrive/SharePoint, you can manually select a file (Word, PPT, PDF, etc.) and translate it into up to 10 languages at once. In other words, if you have an extensive report or presentation that needs complete localization, SharePoint’s translation service can do it. It even handles elements like headers, tables, and layouts – as one Microsoft article notes, “Translation of Word documents preserves headings, text, headers, footers, and other elements.”

This service is uniquely necessary because Copilot/LLM translation is excellent for blocks of unstructured text but is not designed to handle the complex structure of a Word document (TOC, headers, footers, tables, charts, embedded objects) or a complex PDF layout.

• Pay-As-You-Go Cost: Document translation is a metered service. The published pricing is about $15 USD per 1 million characters. Microsoft offers limited free usage (for example, up to 1 M characters per month during the trial), but beyond that, you pay per character (the Azure-linked subscription is charged). In one real-world test, translating several large files (~10 million characters) cost roughly $126,15. By contrast, AutoFill is $0.005 per page – much cheaper for metadata tasks. This highlights how expensive full-document translation can be if your content is heavy on text.
• Processing Time and Limits: Translating large files can take minutes (ten minutes for a 30 MB Word file was reported). There are size limits (e.g., 40 MB max), and only supported file types can be translated. Also, encrypted or password-protected files won’t be translated (the service skips them).
• Quality Caveats: The automated translator generally does a decent job, but it isn’t perfect. Testers have seen formatting quirks – extra line breaks or spaces when the translated text is longer, requiring manual cleanup. A human to verify a translated document is still a must – IMHO.

Comparing Approaches

In effect, for nearly all everyday needs – metadata tagging, straightforward page content – Copilot and autofill have you covered. The only gap is translating entire documents or transcripts while maintaining the layout.

Conclusion: Dead… or Just Sleeping?

So, is the translation service dead? Mostly. For all the small‐scale AI tasks in SharePoint, it’s obsolete. Thanks to AutoFill and Copilot, you rarely need the old translator except when you have a whole file to localize. Microsoft itself seems to acknowledge this: the SharePoint document translation is a standalone Syntex (pay-per-use) feature, whereas Copilot’s abilities overlap heavily with it.

In short, Translation is effectively dead for Metadata/Pages/Content, but it’s still alive for complete documents. It would streamline the AI workflow if Microsoft included document translation in Copilot (or made it broadly free). That way, the only remaining use case for a separate translator – end-to-end file localization – could be handled without extra cost. Until then, the translator service survives only for the big jobs.

Hope that helps,

Renewed Revolution!