Version 4.0

  • Published on Nov 26, 2025
  • 2 minute(s) read
  • eA
 Prev Next

Version 4.0

26 November, 2025

Enhancements

VA

  • "Knowledge Retrieval" Tool Type for AI Agents:
    AI Agents now support a new "Knowledge Retrieval" tool type, enabling direct integration with knowledge bases to search and retrieve relevant information. This enhancement reduces integration effort and improves response accuracy.

  • Knowledge Base Developments:

    • Retrieve Training Status of Documents and Websites in a Knowledge Base:
      Users can now monitor the real-time training status of documents and websites, providing complete visibility into processing progress. Training stages are automatically polled and updated on the UI without requiring a page refresh, displaying detailed progress for document processing and website crawling. Training statuses transition through stages, including "Processing," "Indexed," and "Failed," with automatic error logging for troubleshooting. A "Refresh Index" button becomes active when failed sources are detected, enabling users to retry the training process for failed resources.

    • Delete Data Source from Knowledge Base:
      Users can now remove unwanted documents and websites directly from a knowledge base, providing better control over knowledge base content. Each data source displays a delete icon that triggers a confirmation dialog before removal, ensuring intentional deletions. Deletion is prevented during active training to maintain data integrity.

    • Knowledge Base Versioning:
      Knowledge bases now support versioning through project publishing, enabling the controlled deployment of content across project versions. When a project is published, all knowledge bases are automatically cloned to the target version, with real-time progress tracking through dedicated status indicators including "Cloning," "Ready," "Not Ready," and "Cloning Failed." The system ensures data integrity by requiring all knowledge bases to be in "Ready" status before allowing project publication.

    • Knowledge Base File Size Limits per Tenant:
      Tenant-specific file size limits can now be configured for knowledge base content, ensuring optimal resource allocation and efficient storage management.

  • Ask LLM Node:
    A new "Ask LLM" node has been introduced to the node palette, enabling flow designers to send chat completion requests directly to their preferred LLM provider without requiring knowledge bases or external URLs. This node leverages preconfigured LLM configurations, eliminating the need to enter API keys for each request manually and streamlining the design process.

  • Customer Rating Trends Graph on Dashboard:
    A new "Customer Rating Trends" graph has been added to the dashboard, displaying rating distribution and trends over time. This visualization provides insights into customer satisfaction and helps identify areas for service improvement.

Improvements

VA

  • Tagalog Language Support:
    Tagalog language support has been added across NLU, text normalization, and sentence embedding modules, expanding the platform's multilingual capabilities.

  • Sentiment Graphics Color Adjustment on Dashboard:
    The color scheme for sentiment graphics on the dashboard has been updated for better clarity and consistency.

  • Session Count Alignment Across Portal, Dashboard, and Conversation Pages:
    Session counts are now consistently displayed across the portal, dashboard, and conversation pages through improved license validation and tracking mechanisms.

  • ELK Field Structure Optimization:
    Redundant and duplicate fields in ELK logs have been reorganized to improve data clarity, query performance, and overall system observability.

  • Minor UI Improvements:
    Several minor improvements have been made to the user interface to enhance user experience and usability.

  • Minor Bug Fixes:
    Various minor bugs have been addressed to improve overall system stability and performance.