9 Oct 2026
6 min read
Multilingual Umbraco sites, without the translation bottleneck
How AI and Jumoo’s Translation Manager turn translating an Umbraco site from a slow export-and-email process into a pipeline that runs inside the backoffice.
Umbraco has handled multilingual content well for years. Language variants let one page carry every language side by side, and editors can switch between them in a click. The hard part has never been storing translations. It’s getting them done: exporting copy, emailing spreadsheets to agencies, waiting, pasting the results back in and checking nothing broke on the way.
AI has changed the economics of that process. Combined with the right tooling inside Umbraco, a translation that used to take weeks can be drafted in minutes, leaving people to spend their time on the parts that need judgement.
The old pipeline
On most multilingual sites we inherit, translation works like this: someone copies content into a document, sends it to a translator, gets it back days later and re-enters it page by page. Every step loses context. Translators see words without layouts. Editors re-key text into block lists and nested content. Updates to the English page don’t reach the other languages until someone notices.
It’s time-consuming, prone to error and, quite frankly, boring. AI can speed up the process, make cost savings and streamline the workflow.
Where Translation Manager comes in
Jumoo’s Translation Manager, built by long-time Umbraco contributor Kevin Jump, moves the whole workflow into the Umbraco backoffice. Editors pick the pages they want translated, choose the target languages and send a job. Translation Manager strips out the configuration and markup and sends only the words that need translating, then puts the results back in the right fields, including inside blocks and nested content.
It connects to the services you already use. Professional agencies can work through industry-standard XLIFF files or direct integrations. Machine translation can come from DeepL, Microsoft or Google. And its AI connectors let you translate with large language models from OpenAI, Google Gemini, Azure or locally hosted models, with control over the model and the prompt.
What AI adds
Keeping people in the loop
Machine output still needs checking, especially for legal copy, regulated products and marketing headlines. Translation Manager handles this well: returned translations can be compared with the source, reviewed and approved page by page before anything is published. We recommend AI for the first pass on everything, and a native speaker’s review for anything customer-facing.
A pipeline we’d recommend
The result
Multilingual stops being a project and becomes part of everyday publishing. New markets launch faster, translation budgets go further, and editors stay in Umbraco instead of juggling spreadsheets.
If you run a multilingual Umbraco site, or want to launch one, we can help you set up Translation Manager and an AI translation pipeline that fits your content and your review process.