Skip to main content

Introducing more powerful Multilingual Support (MLS)

Related products:AI AssistantEnterprise Search
  • March 25, 2026
  • 2 replies
  • 187 views

Penny Zhou

We're excited to announce more powerful, upgraded Multilingual Support (MLS) capabilities coming to Assistant and Enterprise Search!

Today, via the Knowledge Search (KS) plugin, Moveworks supports non-English search for Knowledge Base Articles (KBAs) by converting both non-English queries and non-English KBAs to English in order to process them. While this approach provided multi-language support, it introduced risk of accuracy loss along the way. Additionally, non-English search is currently not supported for files.

With our upgraded, native MLS capabilities, we will support non-English search for files along with KBAs. We’ll eliminate the "double translation" approach entirely, leading to improved accuracy and reliability. Our new upgraded search capabilities, powered by next-generation retrieval and ranking models, work directly over content in its original language across 30+ languages, delivering better relevance quality and more reliable results.

We'll also simplify language and geo controls with this upgrade. Today, admins must manually configure language and geo boost settings, and users must explicitly set a preferred language for it to influence results. With enhanced MLS and the new Enterprise Search (ES) plugin (more details here), both become native ranking signals in the core stack, which means employees will automatically see results best matched to their locale and query language without requiring manual configurations. You will still be able to interact with Assistant in 100+ languages as you do today.

In late Q2, we'll further enhance user control by replacing today's "preferred bot response language" with two settings that apply across Assistant and Search:

  • App Interface Language: Sets the language of the UI (navigation, controls, labels), independent of what language a user types.
  • Response Language: Sets the default language for responses, employee communications and notifications. Regardless of this setting, users can interact in any language and the experience will respond in the language used for that interaction.

We'll also display citations in the source’s original language, with the option for users to translate them into their query language when needed. 

Refer to the table below for a comparison of current and future state.

When will this MLS upgrade be available?

Early access to new MLS capabilities will be available starting early Q2 onward, with general availability starting mid-late Q2.

Get early access

If you'd like to try upgraded MLS capabilities ahead of the general rollout, please fill out this form to register your interest for early access. Reach out to your account team with any questions.

Functionality Current State:
KS plugin
Future State:
ES plugin + MLS
Non-English content search KBAs: supported ✅
Files: not supported ❌
KBAs: supported ✅
Files: supported ✅
Admin controls on geo/language relevance Three admin-configured controls for geo + language Admin boost controls removed; geo/language handled natively via ranking signals
Geo handling Two admin controls: Geocode Boost (KBAs) + File Search Geocode Boost (Files) with Disabled/Soft/Hard Geo becomes a ranking signal (no hard/soft settings)
Language handling One admin control: Language Boost with Disabled/Soft/Hard Language becomes a ranking signal (no hard/soft settings)
Retrieval and ranking language handling Double translation of non-English queries and non-English documents Retrieval and ranking done natively in original languages
Query/response language handling Assistant understands all languages, but will respond in the bot preferred response language if it’s set

Scenarios:
• English users can only see English sources in the response, no matter what settings
• Spanish users may see English and/or Spanish sources in the response, depending on the setting
Assistant understands all languages and will respond dynamically in the user’s query language regardless of user-set response language

Scenarios:
• English users may see Spanish (or Chinese) sources in the response
• Spanish users may see German (or Chinese) sources as part of the citations in the response
User preferred language (coming in late Q2) Users can set their own bot preferred language through Assistant; admin-default used if not set User can set two language settings: UI and preferred response; admin-default used if not set
Citations display language (coming in late Q2) Citations are shown in the bot preferred language, either natively or translated on-the-fly Citations are shown in the original content language, with an option for the user to translate to the query language

2 replies

Forum|alt.badge.img
  • Community Manager
  • March 30, 2026

This is an exciting update!

The move toward native ranking and eliminating "double translation" is a massive win for accuracy and reliability. As we prepare our partner enablement materials for the Early Q2 rollout, we’d love to clarify a few points to help our customers build the best possible content strategy.

Implementation & Content Strategy

  • 1. Content Architecture: Since we now support native ranking, what is the recommended approach for customers with existing localized content?

    • Do we suggest a "Master Article" approach (one language) vs. separate localized articles in the source?

    • How does the ranking engine handle Mixed-Language pages (e.g., a single KBA or File containing both English and Japanese instructions)?

    • Does the system have a preferred source language for the highest retrieval accuracy, or is performance uniform across all supported language pairs?

  • 2. ServiceNow (NOW) Integration: How should customers look to balance ServiceNow’s Localization Framework with our native ranking to ensure the most efficient content management process?

  • 3. Attribute Personalization: Are there plans to expand attribute-based personalization for search responses (e.g., by Role or Department) beyond the current Location-based signals?


Penny Zhou
  • Author
  • Community Manager
  • April 7, 2026

Rollout update —MLS capabilities will roll out in two stages. 

Stage 1: MLS for KBAs and FAQs

MLS enhancements for KBAs and FAQs will be bundled with the ES plugin update. Once live, the upgrades (rows 2–6 in the table above) will automatically apply to all your existing KBAs and FAQs. To see when you will receive the updated ES plugin, refer to this post for the rollout schedule.

Stage 2: MLS for Files — May 2026

For the first time, AI Assistant and Enterprise Search will support non-English search across files (PDFs, Word documents, PowerPoints, and text files)! This update extends the same intelligent, language-aware ranking available for KBAs to all file types, so employees get relevant answers across all content. File results will appear alongside KBA results and automatically upranked based on the user's language and locale. To prepare, work with your account team to ingest non-English files into Moveworks ahead of the rollout.

Early access: Thank you to everyone who expressed interest in the LP program! As a part of the LP program, you will be able to test MLS for file search in your sandbox environment before general availability in May. We will reach out to confirm your participation later this week.