Version 4.6

  • Updated on Mar 16, 2026
  • Published on Mar 16, 2026
  • 2 minute(s) read
  • eA
 Prev Next

Version 4.6

16 March, 2026

Enhancements

VA

  • LLM Configuration Health Check on Save:
    When saving an LLM configuration, the system now automatically performs a health check to validate the provided credentials and verify connectivity. This allows users to instantly confirm that their configuration is correct without manual testing.

  • Context Variables in RAG Agent & Ask LLM Instructions:
    Variables can now be used within the Instructions field of RAG Agent and Ask LLM nodes, enabling the creation of reusable instruction templates across similar nodes and reducing design effort.

  • ElevenLabs TTS Custom Model Selection:
    Scenario designers can now select from ElevenLabs' available TTS models when configuring text-to-speech settings. This allows choosing the most suitable model based on project requirements or target language.

  • Advanced Conversation Routing (ACR) as a Tenant Feature:
    The Advanced Conversation Routing feature is now managed as a tenant-level feature flag, giving product administrators full control over which customers have access to this capability.

  • AI Agent Dashboard Trend Graph Granularity Control:
    Users can now adjust the time granularity of trend graphs on the AI Agent dashboard, enabling more detailed or broader views depending on their analysis needs.

  • Conversations List Export with Custom Columns:
    The Conversations page export now includes all visible custom columns, such as Rating, in addition to default fields. This ensures exported files accurately reflect the configured view for offline analysis.

  • Ext API for Customer Rating (CSAT) Data:
    A new External API endpoint has been introduced to expose customer rating (CSAT) data from VA sessions. The endpoint returns session count, total rating values, and the number of sessions with a rating of 4 or 5 for a specified date range.

  • Ext API for Negative Sentiment Data:
    A new External API endpoint has been introduced to expose negative sentiment data from VA sessions. The endpoint returns the total number of analyzed sessions alongside the count of sessions with a negative sentiment score for a specified date range.

Improvements

VA

  • Bothub Metadata Update:
    Bothub metadata has been updated to version 4.5.1, ensuring compatibility and incorporating the latest improvements.

  • SMTP CRL Check Disabled:
    The Certificate Revocation List (CRL) check for SMTP connections has been disabled, improving reliability for email integrations in environments where CRL endpoints may be inaccessible.

  • Knowledge Base Timeout Duration for FAQ Documents:
    The ingest request timeout for FAQ-type documents has been extended, reducing processing failures for large knowledge base uploads and ensuring more reliable document ingestion.

  • ACR Prompt Enhancement:

  • Intent classification prompts within the hybrid NLU system have been refined to improve accuracy for complex, ambiguous, or inverted user expressions.

  • 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.