Version 5.31
June 22, 2026
Enhancements
Extended LLM Provider Support
The Generative AI settings page now supports a broader set of large language model providers. AWS Bedrock, Google Gemini, Groq, and Together AI can be configured as the Generative AI backend, allowing organizations to leverage existing provider contracts and infrastructure. LiteLLM is also supported as a proxy layer, enabling teams that manage multiple models through a unified interface. Organizations running self-hosted models can configure a custom vLLM endpoint directly as the Generative AI source.
AI Coaching with Performance Metrics
AI Coaching now incorporates conversation and acoustic performance data into its analysis. Metrics such as average call duration, silence ratio, hold count, agent talk share, speaking pace, and interruption frequency are evaluated alongside coaching insights. Administrators can define target values for each metric from the settings page, enabling data-driven benchmarks for agent performance.
Masking Exception List
An exception list has been added to the masking configuration. Administrators can explicitly specify values that should not be masked, preventing business-critical data from being hidden unintentionally while maintaining privacy controls across the platform.
Conversation Detail: Related Group Identifier
The conversation detail page now displays the related group identifier. When multiple conversations are linked to the same customer interaction, users can identify and navigate between grouped conversations directly from the detail view.
Improvements
AI Coaching Overview
The AI Coaching overview page has been reorganized to present coaching recommendations and performance summaries in a clearer, more accessible layout.
Coaching Recommendation Explainability
Supervisors can now view the specific conversations that each AI Coaching recommendation is based on, directly from the AI Coaching page. This transparency allows supervisors to validate recommendations and review source interactions when needed.
Indexing Isolation for Problematic Tenants
Indexing processes from tenants experiencing failures are now temporarily suspended to prevent them from affecting the indexing performance of other tenants. This change improves overall system stability and processing throughput.
Minor UI Enhancements
Several minor improvements have been implemented to enhance the user interface. These enhancements make the interface more intuitive and user-friendly, contributing to a better overall user experience.
Minor Bug Fixes
Various minor bugs have been addressed to improve overall system stability and performance. These fixes result in a smoother and more reliable user experience.