SUVA v23.11 Release Notes

SUVA’s v23.11 release will be available to all users starting on November 30th. This new release introduces enhanced LLM capabilities, user interface refinements, and analytics improvements.

The highlight of SUVA v23.11 is the reformed LLM integration process, which introduces SUVA Partner Provisioned LLMs, including Claude by AWS.

The new release enhances the SUVA admin panel with improved user experience (UX), making navigation smoother for users.

In addition to the mentioned updates, the accuracy of session tracking has been improved in chatbot interactions.

Highlights of SUVA v23.11 Release

1. SUVA Admin Panel’s Navigation

   If you are familiar with the SUVA admin panel, you will notice some changes upon upgrading your SUVA instance to the new release. Some prominent changes are:

   • Analytics is now available in the left navigation menu.

     

     Earlier, this section was a part of the Virtual Agent Configuration screen.

     

   • Large Language Model has also found a section in the left navigation menu.

     

     Earlier, it was a part of the Manage Virtual Agent screen.

     

   • Intent Suggestions has now been renamed to NLP Manager and can be spotted in the left navigation menu.

     

     Earlier, this was available in the Virtual Agent Configuration screen. You can also see that the Language filter is also missing in the new navigation panel.

     

2. SUVA’s Partner-Provisioned LLMs

   LLM Integration has undergone a complete transformation. Apart from OpenAI, integration is now allowed with SUVA partner-provisioned LLMs like Claude by AWS to provide more accuracy in chatbot responses.

   The partner provisioned LLM also helps ensure data privacy as these models do not store data, unlike other third-party LLMs.

   

   SUVA now allows you to activate more than one LLM at a time. Furthermore, you can link one or more Virtual Agents to an LLM, though each Virtual Agent can be connected to only one LLM.

   Related doc(s): Integrations: Integrate SUVA Agents with Large Language Models

3. Session Tracking Improvements

   Session tracking has been updated with new touchstones to identify the exact number of sessions and conversations in SUVA.

   Among the new criteria for terminating a session, are:

   • A session is terminated when a user logs out on a landing page.

   • An ongoing session is terminated after an hour of inactivity.