SUVA Q2 '24 Release Notes

The second quarterly release of the year (Q2 '24) is now available for all SUVA users. The highlights of this release include Azure OpenAI Service integration support, overhauled Overview section in the Analytics module, and the ability to extend connections between nodes in the storyboard to other stories and LLM responses.

The user interface (UI) of the SUVA chat window has been upgraded and sees new features like Email Chat and Citations in LLM-powered responses.

Overhauled Analytics > Overview

The Analytics module's Overview section has been completely revamped. The new dashboard provides admins with greater visibility into chatbot performance and customer satisfaction.

Azure OpenAI Service Integration

If your organization relies on Azure OpenAI Service for LLM support, then this release brings a good news for you. SUVA now supports Azure OpenAI Service integration in the Bring your own LLM section.

Creating Connected Conversational Experience via Connecting Storyboards

SUVA admins are given the capability to connect nodes further to create a connected experience for your chatbot users. There are three ways to connect nodes:

Storyboard: Connect a node to another storyboard.
LLM Response: Link a node directly to an LLM response.
Continue in Story: Keep the context within the current story.

Connecting storyboard nodes to different stories allows for the creation of more complex, layered, and personalized conversations that can dynamically switch contexts based on user interactions.

Upgraded SUVA Chat Window

The new release introduces noticeable changes to the SUVA chat window, including a new user interface. Below is an illustration of the new chat window:

Related Doc: An Overview of the SUVA Chat Window

Download Chat Transcript

Often, chatbot users want to save responses for future reference. They can now do this by downloading the chat transcript directly from the chat window.

Citations in LLM-Generated Responses

The process of citing sources in LLM-generated responses has been streamlined. Citations now appear as quotation icons within the response. Hovering over these icons reveals the information sources.

Enhancement(s)

Improved LLM Bot Responses with a Better Understanding of User Context

Our goal is to continuously improve the quality of chatbot interactions. To achieve this, we've refined our bots to create more human-like responses by better understanding user context, especially for follow-up queries. This approach prevents the bots from confusing contexts and generating inappropriate responses.
Tracking Handover

There may be instances where a chat user opts to connect with a live agent after failing to find a satisfactory answer through the chatbot. Previously, it was not possible to determine whether the handover was successful or unsuccessful (due to unavailability of an agent or any other reason). Now, a tracking mechanism has been introduced in SUVA to monitor successful handovers with the case IDs.
LLM Token Consumption Details

The Large Language Model > Usage Dashboard previously only showed high-level details like the total number of tokens consumed by LLMs. Now, admins can view details of each bot activity that consumed an LLM token. These details are visible when expanding the Total Requests section in the Usage Dashboard.

Related Doc: Large Language Model (LLM) Usage Dashboard
The SUVA admin panel can now be browsed in languages other than English. Users can access the admin panel languages including, Spanish, German, and French.
In unusual situations, training and publishing a bot could get stuck. Earlier, there was no way to restart training or publishing a bot. A 'Reset' button has been introduced to terminate the ongoing training/publishing and start training and publishing again.

Last updated: Wednesday, September 25, 2024

Or, send us your review at help-feedback@searchunify.com