Multilingual support for ServiceNow Otto (built on Now Assist) and AI Agents

[Ashley Snyder]

UPDATED: 11/24/2025

This article covers commonly asked multilingual questions when using ServiceNow Otto (built on Now Assist) and AI Agents products.

How does ServiceNow Otto support translation across languages?

ServiceNow Otto offers integrated translation capabilities to provide multilingual experiences for AI-driven workflows:

•   ServiceNow Otto Native Translation: This feature leverages the multilingual capabilities of the Now LLM Service, Microsoft Azure OpenAI, Anthropic Claude on AWS, and Google Gemini to directly understand and generate content in multiple languages. It delivers high-quality translation for generative AI content such as summaries, resolution notes, and knowledge generation, and more.

Which languages are supported by ServiceNow Otto?

Languages are categorized by priority, with different translation methods applied.

• P1, P2, and P3 Languages: ServiceNow Otto Native Translation primarily supports these languages using the large language model's (LLM) native multilingual capabilities.
  • P1: English, French, German, Italian, Spanish, Brazilian Portuguese.
  • P2: Canadian French, Japanese, Dutch. 
  • P3: Swedish, Finnish, Czech, Hebrew, Hungarian, Korean, Norwegian, Polish, Portuguese, Russian, Simplified Chinese, Traditional Chinese, Thai, Turkish, Arabic, Danish.

• In addition to P1, P2, and P3, ServiceNow Otto can support other languages, provided they are available from the model providers, such as Microsoft Azure OpenAI. Refer to the model provider model cards for more information.
  • If you need assistance with supporting a language that is not on the list of P1-P3 supported languages, please contact ServiceNow Support.
• Our recommendation is to use Microsoft Azure OpenAI as the selected model provider for your ServiceNow Otto skills or AI Agents when you require support for languages other than English.

However, specific model support varies:

• ServiceNow Large Language Model v2 (GPT-OSS) currently support English, French, German, Spanish and Italian natively.
• ServiceNow Small Language Model v2 (Apriel) supports all P1 & P2 languages.
• Microsoft Azure OpenAI GPT-4.1/5.2 and GPT-4.1/5-mini supports all P1, P2, & P3 languages.
• Google Gemini 2.5 Flash/2.5 Pro supports all P1, P2, & P3 languages, with the exception of French Canadian support natively.
• Anthropic Claude 4.0/4.5 Sonnet and 4.5 Haiku supports all P1, P2, & P3 languages, with the exception of French Canadian and Dutch support natively.

What if I need to translate static text such as forms or live Agent Chat outside of ServiceNow Otto usage?

For non-generative content on the platform, such as translating platform fields, knowledge articles, or live Agent Chat, you can use:

• Platform Dynamic Translation: This is a separately licensed capability that integrates your own translation provider (e.g., Microsoft Azure, Google, IBM) into the Now Platform. This option requires configuration and a customer-managed translation provider.
• Examples of when to use Platform Dynamic Translation:
  • Translating UI Elements: If you need to translate text on portals, forms, static fields, or other user interface components that are not part of a generative AI response.
  • Live Agent Chat: For real-time translation of conversations between agents and end-users in live chat sessions.
  • Knowledge Articles: To translate existing knowledge articles within the platform for non-generative AI use.
  • Email Notifications: To translate email notification templates.

Key Considerations:

• Customer Configuration: You must configure your own translation provider (e.g., Microsoft Azure, Google, IBM) to use this capability.

For more in-depth information on Platform Dynamic Translation, please refer to the product documentation and community articles:

Product documentation – Platform Dynamic Translation

In Platform Language Support Guide

Configure Localization Framework (LF) and Dynamic Translation (DT) on an instance

Need to translate a configurable workspace?

Need to translate a portal?

Can you provide a comparison of the different translation options?

Feature	ServiceNow Otto Native Translation	Platform Dynamic Translation
Use Case	Used across all ServiceNow Otto Skills and AI Agents for understanding and generarting content in multiple languages.	Used for non-generarive AI congent such as forms, portals, static fields, and live Agent Chat.
Language Scope	Supports P1, P2, and P3 languages, see the model provider support above.	Supports all platform-supported languages.
Provider Setup	No configuration required, toggle on and select language in ServiceNow Otto Admin console, leverages multilingual capabilities of Now LLM Service, Microsoft Azure OpenAI, Anthropic Claude on AWS, and Google Gemini.	Requires integration of your own translation provider (Microsoft Azure, Google, IBM).
Entitlement	Included in ServiceNow Otto, no additional entitlements, providers, or assists required.	Requires a separate customer-managed translation provider.

What does an end-to-end multilingual workflow in ServiceNow Otto look like?

We recommend using ServiceNow Otto Native Translation with an appropriate model provider for organizations aiming to provide seamless language support across self-service and AI skills.

Here's an example:

The requester starts in your portal(s):

• A requestor uses ServiceNow Otto for Virtual Agent to ask a question in Spanish.
• ServiceNow Otto Native Translation processes the request and generates a synthesized response in Spanish using the configured model provider's native mutlilingual capabilities. 

Scenario: Cross-lingual Q&A with ServiceNow Otto If the user asks a question in Spanish, for example, "¿Por qué el servidor MID no se inicia después de la instalación?", the system will:

1. Detect the language: The system first detects the user's language based on their language & region settings or the content of the query itself.
2. Retrieve relevant content: It then attempts to retrieve the most relevant content using Retrieval Augmented Generation (RAG) in the user's language (Spanish).
3. Language Fallback: If relevant Spanish Knowledge Base articles are not found, the system will apply language fallback, searching for content in other configured fallback languages, such as English.
4. LLM Generation: The configured model provider (such as Azure OpenAI, Google Gemini, Anthropic Claude, or Now LLM) then processes the retrieved content (in Spanish or English) and generates an answer directly in the user's preferred language (Spanish) using it's native multilingual capabilities.

Escalating to a live agent:

• The requestor escalates to a live agent. At this point, Platform Dynamic Translation translates the conversation between the agent and requestor. This requires prior setup of a translation provider by the admin.
• The agent uses the Chat Summarization skill. If the language is supported by ServiceNow Otto Native Translation and configured, it is processed via the assigned model provider.

Recommendation for a seamless workflow: For this workflow to function seamlessly, customers should:

• Install the appropriate language packs for each translation target language, if applicable. For ServiceNow Otto use cases and languages that do not contain language packs contact ServiceNow Support for assistance.
• Select a model provider for the specified skill that meets your language requirements in the AI Admin Hub.
• Enable ServiceNow Otto Native Translation in the AI Admin Hub.
• Select all relevant P1, P2, and P3 languages needed by their users for the model provider(s).
• Enable Platform Dynamic Translation in their instance if they wish to translate live Agent Chat or other static UI elements.
• Bring and configure their own translation provider (e.g., Azure, Google, IBM) for Platform Dynamic Translation.

This configuration ensures uninterrupted language coverage across Virtual Agent, live agent handoffs, static content, and ServiceNow Otto generative workflows.

Relevant Documentation for End-to-End Workflow Configuration

To configure the various components that comprise the end-to-end multilingual workflow, customers should refer to the following official product documentation:

1. Language pack installation: Fundamental to all multilingual capabilities, ensuring the necessary language resources are available on your instance.

• What to configure: Install the required language packs for all languages you intend to support across any of the translation types.
• Where to find information: This is typically found in general ServiceNow platform administration documentation, often under "System Localization" or "Language Support."

2. General multilingual service configuration for ServiceNow Otto: This is the foundational step for enabling translation capabilities within ServiceNow Otto applications.

• What to configure: Enable ServiceNow Otto Native Translation and select the languages you wish to support. Choose a model provider that supports your required languages. (Microsoft Azure OpenAI is recommended for comprehensive multilingual support).
• Where to find information:
  • Multilingual service for ServiceNow Otto applications: This section within the AI Admin Hub is where you manage these settings directly for both ServiceNow Otto Native Translation.
  • Translation for ServiceNow Otto product documentation: This product documentation covers the setup and management of both Native Translation for ServiceNow Otto applications.

3. ServiceNow Otto Virtual Agent multilingual configuration: This ensures your Virtual Agent can communicate and generate responses in multiple languages.

• What to configure: Configure Native Translation for languages with AI Admin Hub, select the desired model provider for the AI Agents skill group in the AI Admin Hub. This allows the Virtual Agent to leverage ServiceNow Otto's translation capabilities for user interactions and synthesized responses.

4. AI Search and ServiceNow Otto Q&A Multilingual Configuration This is crucial for cross-lingual search capabilities and translated Q&A responses.
   
   • What to configure: Configure language fallback settings to enable cross-lingual search when content in the user's preferred language is not available.
   • Translation for ServiceNow Otto product documentation: This document also covers aspects of how translation works with ServiceNow Otto skills, including Q&A.

5. Platform Dynamic Translation Configuration (for Live Agent Chat and Static UI) This enables translation for non-generative content and live interactions.

• What to configure: Integrate your chosen third-party translation provider (e.g., Microsoft Azure, Google) with the ServiceNow platform and configure it for use in areas like Agent Chat and static UI fields.
• Where to find information:
  • Configuring Dynamic Translation: This is the main documentation for setting up Platform Dynamic Translation across your instance.
  • Configuring Dynamic Translation for Agent Chat: This guide specifically details how to enable translation for live chat conversations.

By consulting these specific documentation resources, customers can effectively configure the various components that contribute to a seamless end-to-end multilingual workflow in ServiceNow.

How does multilingual support work with AI Search and ServiceNow Otto for AI Search?

ServiceNow Otto for AI Search supports multilingual interactions:

• Hybrid Search and Cross-lingual Search: This allows users to receive ServiceNow Otto responses in their preferred language, even if the content originates in a different language. The configured model provider uses its native multilingual capabilities to generate responses in the user's preferred language.

• Language Fallbacks: AI Search respects global and language fallback configurations. For instance, if a Spanish user queries in Spanish, and Spanish KB articles are not found, AI Search can look for relevant English KB articles (if English is set as a fallback) and the model provider will generate a response in the user's preferred language (Spanish).

How do I configure multilingual services for ServiceNow Otto?

1. Install Language Packs: Install the necessary language packs based on your needs.
2. Enable ServiceNow Otto Native Translation:
   • Navigate to the AI Admin Hub, specifically under Multilingual service > Native Translation.
   • Toggle "Native Translation" ON.
   • Click "Edit" and select the installed P1, P2, and P3 languages you want to use with Native Translation.
3. Contact ServiceNow Support for any languages that are not available, if supported by a ServiceNow-integrated model provider.

[Vivek Verma]

Hello @Ashley Snyde, I'm just curious if there is a guide to enable Native Dynamic Translation. It seems the SN document is not updated yet, but I heard Xanadu patch 3 can dynamically transition using native AI not MS OEM.


LINK: Translation for Now Assist

Note: We are already working on the Xanadu Patch 3 version. 

[Ashley Snyder]

@Vivek Verma It should be up to date now: https://www.servicenow.com/docs/bundle/xanadu-intelligent-experiences/page/administer/now-assist-adm... and we have an AI Academy on 12/10 covering this as well.

[Ricardo Reis]

Hello @Ashley Snyder,

We have an Knowledge Article in English and another with the same content in French.

How does Now Assist with Dynamic Translation work if I search for something in German that is related to both articles?

Will both articles be sent in the results?

Thank you in advance.

[Øyvind Fosse]

@Ricardo Reis 
A bit late with the response but to clarify, the default behaviour of AI Search which is responsible for the retrieval part of a Now Assist answer will be to only retrieve KB articles in the session language of the user. So if you have a user who has selected German but articles are in English or French then it will not find anything.

however you can change this with a sys property: glide.ais.global_searchable_filter.kb_knowledge =

Here you can define a filter condition for articles that should be globally searchable. if you for example set it to active=true then all articles will be included. and since AI Search uses a hybrid search of keyword and semantic search then it would likely find both the French and English article. you can change the filter condition in the sys property so if you filter for English only articles then only the English KB articles will be retrieved when the user is searching with session language set to German. 

the native translation will be able to consume the article and respond in German even if the article itself is in English. 

[Øyvind Fosse]

FYI all, I posted about my experiences testing with Native Language in Now Assist:
https://www.servicenow.com/community/now-assist-blog/unlocking-the-benefits-of-native-language-compr...

[Ricardo Reis]

Hello @Øyvind Fosse Thank you very much for your reply. You were very clear and direct.