Engati - User Guide
Generative AI

Enable LLM for Multilingual capabilities

5min

This document provides an overview of the newly enabled setting IS_LLM_MULTILINGUAL_ENABLED in the Gen AI system, detailing its functionality and benefits.

IS_LLM_MULTILINGUAL_ENABLED

Functionality

  • Language Detection: When this setting is enabled, the AI automatically detects the language of the user's query.
  • Translation: Language detection and translation are handled by the GT model (Google Translate) by default. However, when IS_LLM_MULTILINGUAL_ENABLED is set to True, all language-related processing, including language detection and translation, is handled by the LLM model itself, eliminating the need for Google Translate.
  • Response Generation: Once the language is identified, the AI uses the Gemini model to generate responses in the same language as the query.
  • Seamless Multilingual Support: Users can interact with the AI in their preferred language, and the system will ensure the response matches the query's language.
Document image


Use Case

  • Global Accessibility: This setting allows the AI to cater to a diverse user base by supporting multiple languages.
  • Localized Interactions: Businesses can provide AI-driven customer support in various languages without manual intervention for translation.
  • Enhanced User Experience: Users receive personalized, context-aware responses in their native language, improving satisfaction and engagement.

Benefits

  • Improved Usability: Supports seamless communication across language barriers.
  • Efficiency: Reduces the need for third-party translation tools during interactions.
  • Scalability: Enables organizations to expand their reach to international audiences.
  • Cheaper: LLM Translation is much cheaper than Google Translate and gives as good precision as Google Translate

Example Workflow

  1. Query in Native Language: A user asks a question in Hindi: "मुझे 3 दिन की यात्रा के लिए नैनाटल के लिए एक विस्तृत यात्रा कार्यक्रम दें" which translates to “Give me a detailed itinerary for Nainatal for a 3 day trip”
  2. Language Detection: The AI detects the query language as Hindi using the GT model.
  3. Response Generation: The Gemini model generates the response 
  4. Delivery: The user receives the response in Hindi, matching the language of the query.

If you face any issues or queries please reach out to us at [email protected]