GETTING STARTED
Introduction to AgentAssist
Architecture and Configuration
Glossary
FAQ
Release Notes
Recent Updates
Previous Versions

SET UP AgentAssist
AgentAssist Setup Guide
Sign Up for AgentAssist
Conversation Automation
Automations Board
Create a Bot
Add Multiple Bots
Create Use Cases
Overview
FAQs Setup
Create Dialog Tasks
Welcome Event
LLM and Generative AI
Access Custom Data in Agent AI Bot
Search AI
Agent Coaching
Playbook
Customizing the AgentAssist Widget
Configure Auto Summarization with Custom Dialog Tasks
Feedback Settings
Widget Theming (Layout Customization)
Supported Languages

INTEGRATIONS
AgentAssist Integration with Third-Party Applications
Amazon Connect
Amazon Connect with AgentAssist Voice Installation Guide
Amazon Connect with AgentAssist Chat Using Custom CCP
Amazon Connect with AgentAssist – Voice via AWS Third Party Applications
Channels
Voice
Configure Voice Channel
Configure Speech Recognition
Chat
SmartAssist
Configure AgentAssist
Demos
Voice Demo
Chat Demo
Salesforce
OmniChannel
Omni-Channel Configuration - Chat
AgentAssist Configuration - Chat
Genesys CTI
AgentAssist Voice Integration
NICE-CTI
Kore Agent AI with Salesforce NICE CX-CTI
Demos
Voice Demo
Chat Demo
Zendesk
Set up AgentAssist for Zendesk
Genesys
AgentAssist Integration with Genesys - Manual Setup Instructions
Agent AI Integration with Genesys Cloud CX
Set Up AgentAssist for Genesys
Change AgentAssist Bot in Genesys
Integrate Kore.ai Bot on Genesys Cloud CX Messaging
NICE MAX Desktop
Talkdesk
Twilio

AGENT EXPERIENCE
AgentAssist Widget
Bot Override and Automations

ANALYTICS
Conversation Logs
Dashboard
Agent AI ROI Analyzer (Beta)
APIs
Raw Data API v1
Raw Data API v2
Conversation Summary Trigger

3.x Release Notes

This document provides information on the various releases and the corresponding feature updates and enhancements introduced in version 3.x of the AgentAssist application.

V3.7.1 November 18, 2024

Patch Release

Widget Enhancement

Agent AI for Outbound Calls

Agent AI is now available for both inbound and outbound calls.

Integration Enhancement

Default Kore Agent AI Interaction Widget for Genesys

To streamline call handling and improve efficiency, the Agent AI widget now supports Genesys’ default pop-up window. The widget automatically opens upon receiving a call or message.

V3.7.0 November 3, 2024

Minor Release

Widget Enhancements

Extended Language Support for Localization

The Greeting Messages, Widget Content, Layout Customization, and Negative Feedback Reasons now support the following additional languages:

  • HE – Hebrew
  • ZH_TW – Traditional Chinese
  • PT – Portuguese (Brazil)
  • ID – Indonesian
  • RU – Russian
  • TR – Turkish
  • VI – Vietnamese
  • PL – Polish
  • FA – Persian
  • UR – Urdu

Enhanced Access Control for Search AI

Agent AI now allows restricted access for targeted users by configuring up to three Search AI apps with the following scopes: 

  • Internal Answers: Access to internal sources only.
  • External Answers: Access to external sources only.
  • Auto-Suggestions: Access to auto-suggestions (prompts) only.

Configuration rules:

  • Choose either Internal Answers or External Answers with Auto-Suggestions.
  • Internal and External Answers cannot be combined in one app.
  • Auto-Suggestions can be configured for only one Search AI app.
  • Knowledge AI settings are available only when selecting either Internal or External Answers.

Learn more.

Decommissioning of Agent AI Widget

The widget v2 will be decommissioned by the end of November 2024. A banner has been added in the Voice and Chat channels to notify users that Agent AI widget v2 will be decommissioned. Please upgrade to v3 for continued support.

Integration Enhancements

Summary Support for NICE MAX Desktop

Agents using NICE MAX Desktop now receive a concise summary at the end of a chat or voice session. This summary highlights key points and identifies action items, allowing agents to review important information quickly. Learn more.

Multibot Support for Agent AI Voice in Genesys

Previously limited to a single bot and a department-specific approach, the solution has now been upgraded to support multiple bots and a queue-based functionality. This enhancement enables Genesys Cloud CX to offer a more flexible and scalable solution for businesses seeking to leverage AI to improve agent efficiency and customer satisfaction across multiple departments. Learn more.

V3.6.1 October 21, 2024

Patch Release

Enhanced NICE MAX Desktop Integration

The Agent AI integration with NICE CX MAX Desktop now supports passing custom data to the Agent AI Widget and is compatible with XO v11. Learn more.

Enhanced Salesforce-NICE CTI Integration

The Agent AI integration with Salesforce-NICE CTI now supports all Agent AI features, including automation, transcription, intent identification, Agent Coaching, Agent Playbooks, Sentiment Analysis, and end-of-call summaries. This results in improved personalization and operational efficiency.

V3.6.0 September 28, 2024

Minor Release

“Thumbs-down”/Negative Feedback Reasons Comment Box Updated

The placeholder text in the comment box is now labelled “Tell us more…” instead of “Add additional comments here.”

Summarization in Spanish

For Spanish conversations, auto-summarization is now available in the Spanish language.

Retaining Enable/Disable State of Imported Use Cases

Imported use cases now retain their “enable/disable” state from when they were exported.

Error Message Configuration

The error message display pattern in the Agent AI widget has been updated to help agents view and understand issues without unnecessary redundancy.

You can configure error messages during Use Case creation at the bot level. These messages are displayed in the Agent AI widget for invalid user input. The runtime display varies based on whether a single or multiple error messages are triggered.

Single Error Message

When a single error message is triggered during dialog task execution, the Assist tab displays the error message and the number of occurrences. If the same error message is triggered again, the message does not repeat, but the number of occurrences increases.

Multiple Error Message

When multiple (different) error messages are triggered, they are all displayed in the Assist tab, with a count indicating the total number of error messages.

Importing and Exporting Agent AI Configurations to support CI/CD

To address the issue of manually reconfiguring settings after each export and import of bots, we have introduced the export and import capability of Agent AI configurations via an Export API on the XO Platform. This enhancement streamlines configuration management by enabling partial exports, allowing users to select specific configurations for transfer and improving overall efficiency in managing bot settings. Learn more.

Enhanced Raw Data API

The Raw Data API is enhanced to retrieve and store summary feedback, reasons, and comments.

SmartAssist Bots available for Agent AI in Third-Party Agent Desktop

Admins can now seamlessly add the voice automation bots created in SmartAssist to Agent AI in third-party agent desktops using the same credentials.

AgentAssist Chat Integration with ServiceNow

AgentAssist is now integrated with ServiceNow, enabling agents to access AgentAssist features directly within the ServiceNow environment for a seamless chat experience.

 

V3.5.1 September 14, 2024

Patch Release

Interaction Widget URL Parameters Enhancement

The Interaction Widget URL now supports a customizable qDelimiter parameter for extracting QueueIdentifier from Genesys/Unisys queue names. This enhancement allows for flexible queue name parsing and improves multi-bot functionality through additional URL parameters.

Key updates:

  • qDelimiter Parameter: Middleware now accepts qDelimiter parameter in the Interaction Widget URL for extracting the QueueIdentifier from the queue name.
  • Default Behavior: If qDelimiter is not provided, the QueueIdentifier defaults to the full queue name.
  • Additional Parameters: For multibot functionality, include “multibot=true” and “x_metadata” parameters in the Interaction Widget URL.
  • x_metadata Encoding: The “x_metadata” value must be a Base64 encoded string of the specified JSON.

V3.5.0 September 1, 2024

Minor Release

Expandable and Collapsible Table Templates in the Widget

This update allows agents to manage their workspace more efficiently by controlling the display of table templates, leading to a more organized and personalized view of information.

Key updates:

  • Expand/Collapse Functionality: Icons added to toggle table view.
  • Default View: Tables appear expanded by default.
  • User Control: Agents can adjust table visibility as needed.

Enabling Server-Side Triggering of Conversation Summary in Agent AI

Third-party applications can now trigger a conversation summary by sending the “agentAssist.endOfConversation” event to Agent AI with the help of a new API. Learn more.

Introduction of Additional languages for Agent AI

For localization, feedback, and greetings, Agent AI now supports 33 languages, including Indian, Dutch, and other major languages. It significantly broadens Agent AI’s accessibility and user-friendliness across different linguistic markets, enabling more effective communication and interaction with a global user base.

Key updates:

  • Expanded Language Options: Added Indian, Dutch, and other major languages, a total of 33 supported languages.
  • Comprehensive Integration: Widget Settings, Feedback Settings, and Widget Theming.
  • Specific Application Areas: Auto Summarization, Default Widget Language, Various Feedback Types (Dialog Task, Answers, FAQ, Summary), and Widget Personalization Settings.

Multi-bot AgentAssist Configuration in Genesys

The earlier integration of the AgentAssist widget with Genesys supported only a single bot, which could cater to the customer service operation of a single queue or department. AgentAssist now introduces a multi-bot configuration to support multiple queues and departments simultaneously. This change significantly enhances the AgentAssist widget’s capabilities within Genesys, allowing for more comprehensive and specialized customer service management across multiple departments or service lines.

V3.4.1 August 11, 2024

Patch Release

Widget Landing Tab

The location of the Widget Landing Tab settings is changed in this update:

  • Old location: Widget Settings > Agent AI Channel Specific Settings
  • New location: Widget Theming > Layout Customization > Widget Menu Layout.

This change consolidates widget customization options, making it easier for users to configure the landing tab along with other layout settings.

V3.4.0 July 27, 2024

Patch Release

Search AI Configuration Moved from Account to Bot Level

The Search AI Configuration data is now saved at the bot level instead of the account level.

Enhanced Scroll Functionality in the Widget

In the Agent AI widget, the Assist and the My Bot tabs now contain “up” and “down” arrow buttons at the top and the bottom to let you scroll to the first and last messages. These buttons appear after the messages in the Assist tab are large enough to require a scroll bar.

Agent AI Enabled Dialog Tasks for Playbook Step Adherence and Triggers

To improve the usability and relevance of the Playbooks module, Agent AI now restricts the selection of dialog tasks specified only for the Playbook’s designated channel. This dialog task selection applies during trigger point creation for Dynamic Playbooks and Step Adherence within Primary Playbooks.

Feedback Reason and Comment on Summarization

The Agent AI widget’s Summary feature now offers enhanced feedback capabilities for auto-generated summaries. Agents can provide more specific and actionable feedback by selecting a reason from up to five customizable messages and adding optional comments. Administrators can edit these messages, turn the comment box on or off, and choose the feedback language. This improved feedback system allows agents to pinpoint issues with summaries more accurately, contributing to the continuous improvement of the auto-summarization feature’s accuracy and relevance over time. Learn More.

Language-specific Summarization

The Agent AI widget now supports language-specific summarization for third-party agent desktops. Admins can customize the language of automatically generated conversation summaries by selecting the Custom Dialog Task option in the Auto Summarization Model section. After choosing a preferred language, the system displays a configured dialog task for that language. The summary is then generated based on this language-specific task, allowing for more accurate and culturally appropriate summaries across different languages.

View Summary from Any Active Tab

The Agent AI widget now displays summaries more flexibly across all active tabs. When an agent closes a conversation, the summary appears on whichever tab they’re currently viewing, not just the Assist tab. This change supports the widget’s customizable layout feature, allowing agents to see summaries even if they’ve rearranged or removed certain tabs. For example, a summary can now pop up on the Search tab, enhancing the widget’s versatility and ensuring agents always have access to important conversation summaries regardless of their current view. 

Widget Layout Customization

The Agent AI widget now offers channel-specific layout customization in beta. Users can adjust language settings, menu labels, visibility controls for messages, dialog task timelines, and assist message types. The feature also allows customization of widget menu layout, assist action menu, landing tab settings, and “More” menu behavior. This flexibility enables tailoring the widget to specific channel needs and preferences. While the preview function is still in beta and may not fully reflect the final widget experience, it provides a close reference for customization options. Learn More.

Agent AI ROI Analyzer (Beta)

The Agent AI ROI Analyzer (Beta) demonstrates the financial benefits and efficiency gains from using Agent AI. Users can input historical agent data and filter the analysis by date, bots, and channels. The feature includes a Cost Savings View to display saved costs. It also helps quantify the value of Agent AI implementation and is available in the Agent AI Dashboard under the ROI Analyzer (Beta) section. The feature is accessible via Agent AI > Dashboard > Agent AI ROI Analyzer (Beta). Learn More.

Entry/Exit Event Support in Third-Party Integration

Integration with third-party applications lets Agent AI capture and process events like entry, exit, and internalTransfer between agents to ensure smooth transition and improved user experience. Currently, these events are only supported by Genesys.

V3.3.1 July 13, 2024

Patch Release

This update includes bug fixes.

 

V3.3.0 June 29, 2024

Minor Release

This update includes feature enhancements and bug fixes.

Summarization Service Optimization

The Conversation Summary generation process is updated to handle short, long, and abrupt endings of conversations. For long conversations, the maximum limit is set to 8000 tokens (approximately 6000 words).

Improved Widget Search Experience

The widget search bar is improved for a better user experience. It now shows “Ask a question” as placeholder text with a search icon on the right. As users type in the bar, the placeholder disappears, and the search icon changes to a close (X) icon, which makes it easy to clear the text. Learn more.

Dashboard Updates

AgentAssist dashboard is enhanced with new suggestion relevance metrics and reorganized customer inquiry insights.

  • A new Relevant Suggestions tab is added, showing the percentage of suggestions used by agents.
  • The What are your customers looking for section is revised:
    • “Articles” is renamed to “Answers”, displaying SearchAssist results.
    • A new “Others” tab is added for web suggestions and additional data sources.

Learn more.

Attachment Support in Widget

AgentAssist now accepts file or image attachments during live conversations. 

If the file or image is not automatically detected, you can override and manually attach the file. To maintain system efficiency, attachments are limited to one file at a time and must not exceed 25 MB in size. Learn more.

Dark Theme for AgentAssist Widget

Agents can now customize the widget appearance by selecting a theme. To select it, go to the widget’s Settings section and select Dark or Light. Learn more.

In-line Digital Form

AgentAssist has introduced in-line digital forms in the Assist tab, making it easier to collect customer information during live conversations. Configure the forms with dialog tasks to use this feature and select “Open inline form” in the Web/Mobile SDK Form Behavior setting.

In-line form:

Learn more.

Streamlined Conversation Summary Management

Conversation Summary management has been improved to enhance usability and tracking. 

  • When agents submit a summary, it’s automatically saved and linked to its Conversation ID, making it easily accessible in the Conversation Logs screen.

  • For easy identification and comparison, the modified Conversation Summary before submission is marked as “Edited Summary” with an “eye” icon to view the original summary.

  • Each displayed Conversation Summary indicates whether it has been modified.
  • If summary generation fails, the following message is displayed to inform the user: “Error occurred while generating summary”.

Learn more.

Enabling Channel-Specific Automations and Managing Appearance in the Widget Library

AgentAssist lets you configure channel-specific dialog tasks across all communication channels – Chat, Voice, and Email. You can also specify whether these dialog tasks appear in the list or only when searched in the widget’s library (Search tab). These preferences can be set while creating or editing a dialog task and are available under the Triggers & Permissions tab.

When Enabled: When a dialog task is enabled for a specific channel, it is triggered automatically or manually by the agent during interaction with customers on that specific channel.

Launch Through Library

Enabled: The dialog task is listed under all available dialog tasks in the Search tab of the widget.

Disabled: The dialog task doesn’t appear on the list but remains searchable via the “Ask a question” search bar in the Search tab.

Learn more.

AgentAssist Raw Data API v2

This updated version of the Raw Data API offers the following additional conversation data for deeper insights into AgentAssist interactions and conversation flows:

  • Feedback (reason code and feedback text in addition to like/dislike) from the agent.
  • Complete generated answers with feedback
  • Comprehensive agent activity tracking
  • Full end-user utterances
  • Dialog task execution details
  • Agent search bar activity timing

Learn more.

V3.2.1 June 15, 2024

Patch Release

This update includes bug fixes.

V3.2.0 June 1, 2024

Minor Release

This update includes feature enhancements and bug fixes.

Floating Save Button for AgentAssist Settings

In the AgentAssist Settings section, the Save button is now always visible and accessible without the need to scroll through the subsections.

Dashboard Filters Include Email Channel

You can now filter the AgentAssist Dashboard data based on channels. The AgentAssist Dashboard now includes Email data, in addition to Chat and Voice data. Select the Email filter from the All Channels list.

Auto-scroll Agent AI Widget Content

A new Auto-Scroll Widget Content setting is introduced to manage the auto-scroll functionality for the content in the Assist tab. Admins can enable/disable it through the Widget Settings > AgentAssist Settings > AgentAssist Channel Specific Settings page, while the Agents can access it in the AgentAssist widget > Settings tab. Between these two users, the Agents can override the Admin’s selection.

Entity-Based Summarization

Dialog tasks having “message nodes” didn’t let users customize the Conversation Summary. Hence, to let users customize the Conversation Summary, “message nodes” are replaced with “entity nodes”. To utilize this capability, admins must create dialog tasks with entity nodes and select these dialog tasks in the Auto Summarization Model > Custom Dialog Task section.

For Conversation Summary customization, select a dialog task with “entity node” in the Custom Dialog Task field.

Warm Transfer

Agent AI has introduced the Warm Transfer feature to customize the transfer functionality of the Agent AI widget. This feature specifically caters to third-party configurations, where appropriate transfer events are transmitted.

The Warm Transfer feature can be defined as a conference call between two agents and applies to both Voice and Chat channels. During this process, Agent-2 has the ability to modify the Summary Feedback, but the Run, Override, Send, Copy, Terminate, Restart, and List View buttons remain disabled until Agent-1 disconnects from the call. It has the following options to choose from:

  • Scroll up to the First Message
  • Continue from the Last Message 
  • Configure Dialog Task

Negative Feedback Settings

For better processing and accurate identification of reasons for receiving a “thumbs down”, Admins can now configure a list of negative feedback messages for agents to select. They can also include a comment box for agents to add additional comments. Admins decide whether these two options appear in the widget after agents select the “thumbs down” icon at the end of a Dialog Task and after each answer and FAQ. Additionally, Admins can change the language of the feedback messages by selecting one from the available language list.

Feedback Settings in the AgentAssist UI:

 

Feedback Messages and Comment Box in the Widget:

V3.1.1 May 11, 2024

Patch Release

This update includes bug fixes.

V3.1.0 April 27, 2024

Minor Release

This update includes feature enhancements and bug fixes.

Agent AI Widget Updates

No Results Found and No Dialog Tasks Enabled Messages

  • The Agent AI widget now shows the “No Search Result Found. Could you rephrase your question or ask something else?” message when an agent’s search query on the search bar doesn’t show any results.

  • If no dialog tasks are configured for an account, the message “No dialog tasks enabled for library” appears in the Library/Search tab.

Agent AI Configuration with OpenAI GPT-3.5

For enhanced Real-time Coaching and Playbooks performance, the AgentAI configuration is now upgraded from OpenAI GPT 3.0 to OpenAI GPT 3.5. OpenAI GPT 3.5 delivers more accurate and context-aware responses, thereby enhancing Real-time Coaching effectiveness and Playbooks quality.

Feedback Mechanism for Answers

In an effort to constantly improve the accuracy and helpfulness of answers displayed on the AgentAI widget, Kore has introduced a feedback mechanism for agents. Agents can now click the “thumbs up” or “thumbs down” icon displayed with every answer to indicate positive or negative feedback. This feedback data will help train Agent AI to provide more accurate answers.

Override Mode Updates

The following updates are made for the Override Mode of the AgentAI widget:

Override Mode Off

  • When the customer’s input is expected, the field remains grayed out and displays the entity name followed by “waiting….”
  • The “Override” call-to-action (CTA) appears, and agents can click it to override the customer’s utterance and manually populate the entity field.

Override Mode On

  • The green “tick” mark that appeared after receiving the customer’s input is removed.
  • The Entity Display Name appears with a cursor to indicate editing mode.
  • Introduced the Cross and Check icons.
  • The “keyboard” icon is replaced with the “agent” icon to indicate the agent’s input.

Updated Summarization Feature

Agent AI has improved the Summarization feature of its widget. The latest version comes up with improved summary accuracy and advanced handling of various scenarios in conversations between agents and users. Currently, this update is available only for the US PROD environment and supports English. It is not yet available for DE, JP, EU, AU, and IND environments.

Key Highlights

  • Improved Summary Accuracy: The summary accuracy has been significantly improved through the implementation of a new model, and Agent AI is fine-tuning it for specific use cases.
  • Summary for Small Conversations: Agent AI has effectively addressed issues related to agent transfer scenarios between bots and users, ensuring seamless communication and comprehension.
  • Handling Incomplete Summaries: The new model handles incomplete summaries more effectively, providing more comprehensive and useful insights even when the conversation is not fully concluded.
  • Handling Longer Calls: The new model handles longer calls extending up to 20 minutes. The previous model would generate incomplete summaries for such extended conversations.

Agent AI UI Updates

SearchAssist Configuration Page Updates

The SearchAssist Configuration page inside Agent AI has been updated with the following changes:

  1. Stream ID has been replaced with App ID.
  2. A warning is displayed when a user enters incorrect values in the configuration fields –  “Could not establish connection with SearchAssist. Check all parameters and try again. Hint: Ensure that the domain URL does not contain “https://” and does not have a forward slash “/” at the end.
  3. As soon as a user lands on the configuration page, the following hint is displayed: “Ensure that the domain URL does not contain “https://” and does not have a forward slash “/” at the end.” 
  4. If a user enters incorrect values in the configuration page, the following error message is displayed: “Error establishing connection.

Conversation Transfer Experience (Third-Party Agent Desktop)

Agent AI has introduced the Conversation Transfer feature to customize the transfer functionality of the Agent AI widget. This feature is applicable only tofor the third-party configurations where appropriate transfer events are transmitted. The Cold Transfer type applies to all three channels – Chat, Voice, and Email and has the following options to choose from:

  • Scroll up to the First Message
  • Continue from the Last Message
  • Configure Dialog Task

Learn more.

Configure your own Auto Summarization Model (Third-Party Agent Desktop)

Agent AI has introduced the Auto Summarization Model widget to customize the automatic summarization process. You can now select Kore’s default automatic summarization or the “Custom Dialog Task“ option, which lets you select a “dialog task with a single message node” to automatically generate the conversation summary. This is applicable only for third-party agent desktop. Learn more.

The Agent AI widget’s Assist tab now displays an updated Summary that includes the following:

  • A .ai Suggested logo.
  • The summary textbox with a cross icon for clearing the text.
  • Ability to freely edit the summary within the text box.


After submission, the summary appears under the “
.ai CONVERSATION SUMMARY” title with a Copy button at the bottom-right corner.

Agent Coaching Updates

In Agent Coaching > Utterance Trigger section, Anytime in conversation is now a default selection for the “in conversation” field.

Dashboard Updates

Enhanced Agent Feedback Functionality

The AgentAI dashboard has been updated to include a detailed summary of Agent Feedback for dialog tasks, including “Not Helpful” feedback and additional comments. Users can now get a detailed view of the “Not Helpful” feedback with Responses, Additional Comments, and Counts. Learn more.

Average Widget Usage (Third-Party Agent Desktop)

The Dashboard has been updated to include the Average Widget Usage feature. This feature gives a graphical representation of the average amount of time agents spend on the AgentAI widget. Learn more.

V3.0.1 April 13, 2024

Patch Release

This update includes feature enhancements and bug fixes.

Display Trait-based Sub Dialog

The new “trait” identification method improves how Dialog Tasks are selected for agents by showing the most relevant “child” or sub-dialog Tasks based on customer interactions rather than just the general “parent” Dialog Task. To use the “trait” based Dialog Task display, admins must define a custom script logic in the AgentAssist Settings page. Learn more.

 

V3.0 March 30, 2024

Minor Release

This update includes feature enhancements and bug fixes.

AgentAssist v3.0 for Genesys Cloud Desktop

AgentAssist v3.0 is now available for Genesys Cloud Desktop.

Key Updates

  • In AgentAssist, a new “Widget URL” field is added under Chat Channel > Bot Info. This field displays the AgentAssist Widget URL that can be used on the client side to point to the latest version of AgentAssist.
  • To switch to AgentAssist v3.0 in Genesys, you must keep all configurations unchanged except for the AgentAssist URL for new users and the Data Tables for existing users. Copy the “Widget URL” from AgentAssist and paste it into the corresponding fields in Genesys.

Backward Compatibility

  • Existing URLs in Genesys will continue to lead to v2 for existing users.
  • To switch to v3.0, copy the “Widget URL” from AgentAssist and paste it into the Genesys Data Table for existing users.

On this Page

3.x Release Notes

This document provides information on the various releases and the corresponding feature updates and enhancements introduced in version 3.x of the AgentAssist application.

V3.7.1 November 18, 2024

Patch Release

Widget Enhancement

Agent AI for Outbound Calls

Agent AI is now available for both inbound and outbound calls.

Integration Enhancement

Default Kore Agent AI Interaction Widget for Genesys

To streamline call handling and improve efficiency, the Agent AI widget now supports Genesys’ default pop-up window. The widget automatically opens upon receiving a call or message.

V3.7.0 November 3, 2024

Minor Release

Widget Enhancements

Extended Language Support for Localization

The Greeting Messages, Widget Content, Layout Customization, and Negative Feedback Reasons now support the following additional languages:

  • HE – Hebrew
  • ZH_TW – Traditional Chinese
  • PT – Portuguese (Brazil)
  • ID – Indonesian
  • RU – Russian
  • TR – Turkish
  • VI – Vietnamese
  • PL – Polish
  • FA – Persian
  • UR – Urdu

Enhanced Access Control for Search AI

Agent AI now allows restricted access for targeted users by configuring up to three Search AI apps with the following scopes: 

  • Internal Answers: Access to internal sources only.
  • External Answers: Access to external sources only.
  • Auto-Suggestions: Access to auto-suggestions (prompts) only.

Configuration rules:

  • Choose either Internal Answers or External Answers with Auto-Suggestions.
  • Internal and External Answers cannot be combined in one app.
  • Auto-Suggestions can be configured for only one Search AI app.
  • Knowledge AI settings are available only when selecting either Internal or External Answers.

Learn more.

Decommissioning of Agent AI Widget

The widget v2 will be decommissioned by the end of November 2024. A banner has been added in the Voice and Chat channels to notify users that Agent AI widget v2 will be decommissioned. Please upgrade to v3 for continued support.

Integration Enhancements

Summary Support for NICE MAX Desktop

Agents using NICE MAX Desktop now receive a concise summary at the end of a chat or voice session. This summary highlights key points and identifies action items, allowing agents to review important information quickly. Learn more.

Multibot Support for Agent AI Voice in Genesys

Previously limited to a single bot and a department-specific approach, the solution has now been upgraded to support multiple bots and a queue-based functionality. This enhancement enables Genesys Cloud CX to offer a more flexible and scalable solution for businesses seeking to leverage AI to improve agent efficiency and customer satisfaction across multiple departments. Learn more.

V3.6.1 October 21, 2024

Patch Release

Enhanced NICE MAX Desktop Integration

The Agent AI integration with NICE CX MAX Desktop now supports passing custom data to the Agent AI Widget and is compatible with XO v11. Learn more.

Enhanced Salesforce-NICE CTI Integration

The Agent AI integration with Salesforce-NICE CTI now supports all Agent AI features, including automation, transcription, intent identification, Agent Coaching, Agent Playbooks, Sentiment Analysis, and end-of-call summaries. This results in improved personalization and operational efficiency.

V3.6.0 September 28, 2024

Minor Release

“Thumbs-down”/Negative Feedback Reasons Comment Box Updated

The placeholder text in the comment box is now labelled “Tell us more…” instead of “Add additional comments here.”

Summarization in Spanish

For Spanish conversations, auto-summarization is now available in the Spanish language.

Retaining Enable/Disable State of Imported Use Cases

Imported use cases now retain their “enable/disable” state from when they were exported.

Error Message Configuration

The error message display pattern in the Agent AI widget has been updated to help agents view and understand issues without unnecessary redundancy.

You can configure error messages during Use Case creation at the bot level. These messages are displayed in the Agent AI widget for invalid user input. The runtime display varies based on whether a single or multiple error messages are triggered.

Single Error Message

When a single error message is triggered during dialog task execution, the Assist tab displays the error message and the number of occurrences. If the same error message is triggered again, the message does not repeat, but the number of occurrences increases.

Multiple Error Message

When multiple (different) error messages are triggered, they are all displayed in the Assist tab, with a count indicating the total number of error messages.

Importing and Exporting Agent AI Configurations to support CI/CD

To address the issue of manually reconfiguring settings after each export and import of bots, we have introduced the export and import capability of Agent AI configurations via an Export API on the XO Platform. This enhancement streamlines configuration management by enabling partial exports, allowing users to select specific configurations for transfer and improving overall efficiency in managing bot settings. Learn more.

Enhanced Raw Data API

The Raw Data API is enhanced to retrieve and store summary feedback, reasons, and comments.

SmartAssist Bots available for Agent AI in Third-Party Agent Desktop

Admins can now seamlessly add the voice automation bots created in SmartAssist to Agent AI in third-party agent desktops using the same credentials.

AgentAssist Chat Integration with ServiceNow

AgentAssist is now integrated with ServiceNow, enabling agents to access AgentAssist features directly within the ServiceNow environment for a seamless chat experience.

 

V3.5.1 September 14, 2024

Patch Release

Interaction Widget URL Parameters Enhancement

The Interaction Widget URL now supports a customizable qDelimiter parameter for extracting QueueIdentifier from Genesys/Unisys queue names. This enhancement allows for flexible queue name parsing and improves multi-bot functionality through additional URL parameters.

Key updates:

  • qDelimiter Parameter: Middleware now accepts qDelimiter parameter in the Interaction Widget URL for extracting the QueueIdentifier from the queue name.
  • Default Behavior: If qDelimiter is not provided, the QueueIdentifier defaults to the full queue name.
  • Additional Parameters: For multibot functionality, include “multibot=true” and “x_metadata” parameters in the Interaction Widget URL.
  • x_metadata Encoding: The “x_metadata” value must be a Base64 encoded string of the specified JSON.

V3.5.0 September 1, 2024

Minor Release

Expandable and Collapsible Table Templates in the Widget

This update allows agents to manage their workspace more efficiently by controlling the display of table templates, leading to a more organized and personalized view of information.

Key updates:

  • Expand/Collapse Functionality: Icons added to toggle table view.
  • Default View: Tables appear expanded by default.
  • User Control: Agents can adjust table visibility as needed.

Enabling Server-Side Triggering of Conversation Summary in Agent AI

Third-party applications can now trigger a conversation summary by sending the “agentAssist.endOfConversation” event to Agent AI with the help of a new API. Learn more.

Introduction of Additional languages for Agent AI

For localization, feedback, and greetings, Agent AI now supports 33 languages, including Indian, Dutch, and other major languages. It significantly broadens Agent AI’s accessibility and user-friendliness across different linguistic markets, enabling more effective communication and interaction with a global user base.

Key updates:

  • Expanded Language Options: Added Indian, Dutch, and other major languages, a total of 33 supported languages.
  • Comprehensive Integration: Widget Settings, Feedback Settings, and Widget Theming.
  • Specific Application Areas: Auto Summarization, Default Widget Language, Various Feedback Types (Dialog Task, Answers, FAQ, Summary), and Widget Personalization Settings.

Multi-bot AgentAssist Configuration in Genesys

The earlier integration of the AgentAssist widget with Genesys supported only a single bot, which could cater to the customer service operation of a single queue or department. AgentAssist now introduces a multi-bot configuration to support multiple queues and departments simultaneously. This change significantly enhances the AgentAssist widget’s capabilities within Genesys, allowing for more comprehensive and specialized customer service management across multiple departments or service lines.

V3.4.1 August 11, 2024

Patch Release

Widget Landing Tab

The location of the Widget Landing Tab settings is changed in this update:

  • Old location: Widget Settings > Agent AI Channel Specific Settings
  • New location: Widget Theming > Layout Customization > Widget Menu Layout.

This change consolidates widget customization options, making it easier for users to configure the landing tab along with other layout settings.

V3.4.0 July 27, 2024

Patch Release

Search AI Configuration Moved from Account to Bot Level

The Search AI Configuration data is now saved at the bot level instead of the account level.

Enhanced Scroll Functionality in the Widget

In the Agent AI widget, the Assist and the My Bot tabs now contain “up” and “down” arrow buttons at the top and the bottom to let you scroll to the first and last messages. These buttons appear after the messages in the Assist tab are large enough to require a scroll bar.

Agent AI Enabled Dialog Tasks for Playbook Step Adherence and Triggers

To improve the usability and relevance of the Playbooks module, Agent AI now restricts the selection of dialog tasks specified only for the Playbook’s designated channel. This dialog task selection applies during trigger point creation for Dynamic Playbooks and Step Adherence within Primary Playbooks.

Feedback Reason and Comment on Summarization

The Agent AI widget’s Summary feature now offers enhanced feedback capabilities for auto-generated summaries. Agents can provide more specific and actionable feedback by selecting a reason from up to five customizable messages and adding optional comments. Administrators can edit these messages, turn the comment box on or off, and choose the feedback language. This improved feedback system allows agents to pinpoint issues with summaries more accurately, contributing to the continuous improvement of the auto-summarization feature’s accuracy and relevance over time. Learn More.

Language-specific Summarization

The Agent AI widget now supports language-specific summarization for third-party agent desktops. Admins can customize the language of automatically generated conversation summaries by selecting the Custom Dialog Task option in the Auto Summarization Model section. After choosing a preferred language, the system displays a configured dialog task for that language. The summary is then generated based on this language-specific task, allowing for more accurate and culturally appropriate summaries across different languages.

View Summary from Any Active Tab

The Agent AI widget now displays summaries more flexibly across all active tabs. When an agent closes a conversation, the summary appears on whichever tab they’re currently viewing, not just the Assist tab. This change supports the widget’s customizable layout feature, allowing agents to see summaries even if they’ve rearranged or removed certain tabs. For example, a summary can now pop up on the Search tab, enhancing the widget’s versatility and ensuring agents always have access to important conversation summaries regardless of their current view. 

Widget Layout Customization

The Agent AI widget now offers channel-specific layout customization in beta. Users can adjust language settings, menu labels, visibility controls for messages, dialog task timelines, and assist message types. The feature also allows customization of widget menu layout, assist action menu, landing tab settings, and “More” menu behavior. This flexibility enables tailoring the widget to specific channel needs and preferences. While the preview function is still in beta and may not fully reflect the final widget experience, it provides a close reference for customization options. Learn More.

Agent AI ROI Analyzer (Beta)

The Agent AI ROI Analyzer (Beta) demonstrates the financial benefits and efficiency gains from using Agent AI. Users can input historical agent data and filter the analysis by date, bots, and channels. The feature includes a Cost Savings View to display saved costs. It also helps quantify the value of Agent AI implementation and is available in the Agent AI Dashboard under the ROI Analyzer (Beta) section. The feature is accessible via Agent AI > Dashboard > Agent AI ROI Analyzer (Beta). Learn More.

Entry/Exit Event Support in Third-Party Integration

Integration with third-party applications lets Agent AI capture and process events like entry, exit, and internalTransfer between agents to ensure smooth transition and improved user experience. Currently, these events are only supported by Genesys.

V3.3.1 July 13, 2024

Patch Release

This update includes bug fixes.

 

V3.3.0 June 29, 2024

Minor Release

This update includes feature enhancements and bug fixes.

Summarization Service Optimization

The Conversation Summary generation process is updated to handle short, long, and abrupt endings of conversations. For long conversations, the maximum limit is set to 8000 tokens (approximately 6000 words).

Improved Widget Search Experience

The widget search bar is improved for a better user experience. It now shows “Ask a question” as placeholder text with a search icon on the right. As users type in the bar, the placeholder disappears, and the search icon changes to a close (X) icon, which makes it easy to clear the text. Learn more.

Dashboard Updates

AgentAssist dashboard is enhanced with new suggestion relevance metrics and reorganized customer inquiry insights.

  • A new Relevant Suggestions tab is added, showing the percentage of suggestions used by agents.
  • The What are your customers looking for section is revised:
    • “Articles” is renamed to “Answers”, displaying SearchAssist results.
    • A new “Others” tab is added for web suggestions and additional data sources.

Learn more.

Attachment Support in Widget

AgentAssist now accepts file or image attachments during live conversations. 

If the file or image is not automatically detected, you can override and manually attach the file. To maintain system efficiency, attachments are limited to one file at a time and must not exceed 25 MB in size. Learn more.

Dark Theme for AgentAssist Widget

Agents can now customize the widget appearance by selecting a theme. To select it, go to the widget’s Settings section and select Dark or Light. Learn more.

In-line Digital Form

AgentAssist has introduced in-line digital forms in the Assist tab, making it easier to collect customer information during live conversations. Configure the forms with dialog tasks to use this feature and select “Open inline form” in the Web/Mobile SDK Form Behavior setting.

In-line form:

Learn more.

Streamlined Conversation Summary Management

Conversation Summary management has been improved to enhance usability and tracking. 

  • When agents submit a summary, it’s automatically saved and linked to its Conversation ID, making it easily accessible in the Conversation Logs screen.

  • For easy identification and comparison, the modified Conversation Summary before submission is marked as “Edited Summary” with an “eye” icon to view the original summary.

  • Each displayed Conversation Summary indicates whether it has been modified.
  • If summary generation fails, the following message is displayed to inform the user: “Error occurred while generating summary”.

Learn more.

Enabling Channel-Specific Automations and Managing Appearance in the Widget Library

AgentAssist lets you configure channel-specific dialog tasks across all communication channels – Chat, Voice, and Email. You can also specify whether these dialog tasks appear in the list or only when searched in the widget’s library (Search tab). These preferences can be set while creating or editing a dialog task and are available under the Triggers & Permissions tab.

When Enabled: When a dialog task is enabled for a specific channel, it is triggered automatically or manually by the agent during interaction with customers on that specific channel.

Launch Through Library

Enabled: The dialog task is listed under all available dialog tasks in the Search tab of the widget.

Disabled: The dialog task doesn’t appear on the list but remains searchable via the “Ask a question” search bar in the Search tab.

Learn more.

AgentAssist Raw Data API v2

This updated version of the Raw Data API offers the following additional conversation data for deeper insights into AgentAssist interactions and conversation flows:

  • Feedback (reason code and feedback text in addition to like/dislike) from the agent.
  • Complete generated answers with feedback
  • Comprehensive agent activity tracking
  • Full end-user utterances
  • Dialog task execution details
  • Agent search bar activity timing

Learn more.

V3.2.1 June 15, 2024

Patch Release

This update includes bug fixes.

V3.2.0 June 1, 2024

Minor Release

This update includes feature enhancements and bug fixes.

Floating Save Button for AgentAssist Settings

In the AgentAssist Settings section, the Save button is now always visible and accessible without the need to scroll through the subsections.

Dashboard Filters Include Email Channel

You can now filter the AgentAssist Dashboard data based on channels. The AgentAssist Dashboard now includes Email data, in addition to Chat and Voice data. Select the Email filter from the All Channels list.

Auto-scroll Agent AI Widget Content

A new Auto-Scroll Widget Content setting is introduced to manage the auto-scroll functionality for the content in the Assist tab. Admins can enable/disable it through the Widget Settings > AgentAssist Settings > AgentAssist Channel Specific Settings page, while the Agents can access it in the AgentAssist widget > Settings tab. Between these two users, the Agents can override the Admin’s selection.

Entity-Based Summarization

Dialog tasks having “message nodes” didn’t let users customize the Conversation Summary. Hence, to let users customize the Conversation Summary, “message nodes” are replaced with “entity nodes”. To utilize this capability, admins must create dialog tasks with entity nodes and select these dialog tasks in the Auto Summarization Model > Custom Dialog Task section.

For Conversation Summary customization, select a dialog task with “entity node” in the Custom Dialog Task field.

Warm Transfer

Agent AI has introduced the Warm Transfer feature to customize the transfer functionality of the Agent AI widget. This feature specifically caters to third-party configurations, where appropriate transfer events are transmitted.

The Warm Transfer feature can be defined as a conference call between two agents and applies to both Voice and Chat channels. During this process, Agent-2 has the ability to modify the Summary Feedback, but the Run, Override, Send, Copy, Terminate, Restart, and List View buttons remain disabled until Agent-1 disconnects from the call. It has the following options to choose from:

  • Scroll up to the First Message
  • Continue from the Last Message 
  • Configure Dialog Task

Negative Feedback Settings

For better processing and accurate identification of reasons for receiving a “thumbs down”, Admins can now configure a list of negative feedback messages for agents to select. They can also include a comment box for agents to add additional comments. Admins decide whether these two options appear in the widget after agents select the “thumbs down” icon at the end of a Dialog Task and after each answer and FAQ. Additionally, Admins can change the language of the feedback messages by selecting one from the available language list.

Feedback Settings in the AgentAssist UI:

 

Feedback Messages and Comment Box in the Widget:

V3.1.1 May 11, 2024

Patch Release

This update includes bug fixes.

V3.1.0 April 27, 2024

Minor Release

This update includes feature enhancements and bug fixes.

Agent AI Widget Updates

No Results Found and No Dialog Tasks Enabled Messages

  • The Agent AI widget now shows the “No Search Result Found. Could you rephrase your question or ask something else?” message when an agent’s search query on the search bar doesn’t show any results.

  • If no dialog tasks are configured for an account, the message “No dialog tasks enabled for library” appears in the Library/Search tab.

Agent AI Configuration with OpenAI GPT-3.5

For enhanced Real-time Coaching and Playbooks performance, the AgentAI configuration is now upgraded from OpenAI GPT 3.0 to OpenAI GPT 3.5. OpenAI GPT 3.5 delivers more accurate and context-aware responses, thereby enhancing Real-time Coaching effectiveness and Playbooks quality.

Feedback Mechanism for Answers

In an effort to constantly improve the accuracy and helpfulness of answers displayed on the AgentAI widget, Kore has introduced a feedback mechanism for agents. Agents can now click the “thumbs up” or “thumbs down” icon displayed with every answer to indicate positive or negative feedback. This feedback data will help train Agent AI to provide more accurate answers.

Override Mode Updates

The following updates are made for the Override Mode of the AgentAI widget:

Override Mode Off

  • When the customer’s input is expected, the field remains grayed out and displays the entity name followed by “waiting….”
  • The “Override” call-to-action (CTA) appears, and agents can click it to override the customer’s utterance and manually populate the entity field.

Override Mode On

  • The green “tick” mark that appeared after receiving the customer’s input is removed.
  • The Entity Display Name appears with a cursor to indicate editing mode.
  • Introduced the Cross and Check icons.
  • The “keyboard” icon is replaced with the “agent” icon to indicate the agent’s input.

Updated Summarization Feature

Agent AI has improved the Summarization feature of its widget. The latest version comes up with improved summary accuracy and advanced handling of various scenarios in conversations between agents and users. Currently, this update is available only for the US PROD environment and supports English. It is not yet available for DE, JP, EU, AU, and IND environments.

Key Highlights

  • Improved Summary Accuracy: The summary accuracy has been significantly improved through the implementation of a new model, and Agent AI is fine-tuning it for specific use cases.
  • Summary for Small Conversations: Agent AI has effectively addressed issues related to agent transfer scenarios between bots and users, ensuring seamless communication and comprehension.
  • Handling Incomplete Summaries: The new model handles incomplete summaries more effectively, providing more comprehensive and useful insights even when the conversation is not fully concluded.
  • Handling Longer Calls: The new model handles longer calls extending up to 20 minutes. The previous model would generate incomplete summaries for such extended conversations.

Agent AI UI Updates

SearchAssist Configuration Page Updates

The SearchAssist Configuration page inside Agent AI has been updated with the following changes:

  1. Stream ID has been replaced with App ID.
  2. A warning is displayed when a user enters incorrect values in the configuration fields –  “Could not establish connection with SearchAssist. Check all parameters and try again. Hint: Ensure that the domain URL does not contain “https://” and does not have a forward slash “/” at the end.
  3. As soon as a user lands on the configuration page, the following hint is displayed: “Ensure that the domain URL does not contain “https://” and does not have a forward slash “/” at the end.” 
  4. If a user enters incorrect values in the configuration page, the following error message is displayed: “Error establishing connection.

Conversation Transfer Experience (Third-Party Agent Desktop)

Agent AI has introduced the Conversation Transfer feature to customize the transfer functionality of the Agent AI widget. This feature is applicable only tofor the third-party configurations where appropriate transfer events are transmitted. The Cold Transfer type applies to all three channels – Chat, Voice, and Email and has the following options to choose from:

  • Scroll up to the First Message
  • Continue from the Last Message
  • Configure Dialog Task

Learn more.

Configure your own Auto Summarization Model (Third-Party Agent Desktop)

Agent AI has introduced the Auto Summarization Model widget to customize the automatic summarization process. You can now select Kore’s default automatic summarization or the “Custom Dialog Task“ option, which lets you select a “dialog task with a single message node” to automatically generate the conversation summary. This is applicable only for third-party agent desktop. Learn more.

The Agent AI widget’s Assist tab now displays an updated Summary that includes the following:

  • A .ai Suggested logo.
  • The summary textbox with a cross icon for clearing the text.
  • Ability to freely edit the summary within the text box.


After submission, the summary appears under the “
.ai CONVERSATION SUMMARY” title with a Copy button at the bottom-right corner.

Agent Coaching Updates

In Agent Coaching > Utterance Trigger section, Anytime in conversation is now a default selection for the “in conversation” field.

Dashboard Updates

Enhanced Agent Feedback Functionality

The AgentAI dashboard has been updated to include a detailed summary of Agent Feedback for dialog tasks, including “Not Helpful” feedback and additional comments. Users can now get a detailed view of the “Not Helpful” feedback with Responses, Additional Comments, and Counts. Learn more.

Average Widget Usage (Third-Party Agent Desktop)

The Dashboard has been updated to include the Average Widget Usage feature. This feature gives a graphical representation of the average amount of time agents spend on the AgentAI widget. Learn more.

V3.0.1 April 13, 2024

Patch Release

This update includes feature enhancements and bug fixes.

Display Trait-based Sub Dialog

The new “trait” identification method improves how Dialog Tasks are selected for agents by showing the most relevant “child” or sub-dialog Tasks based on customer interactions rather than just the general “parent” Dialog Task. To use the “trait” based Dialog Task display, admins must define a custom script logic in the AgentAssist Settings page. Learn more.

 

V3.0 March 30, 2024

Minor Release

This update includes feature enhancements and bug fixes.

AgentAssist v3.0 for Genesys Cloud Desktop

AgentAssist v3.0 is now available for Genesys Cloud Desktop.

Key Updates

  • In AgentAssist, a new “Widget URL” field is added under Chat Channel > Bot Info. This field displays the AgentAssist Widget URL that can be used on the client side to point to the latest version of AgentAssist.
  • To switch to AgentAssist v3.0 in Genesys, you must keep all configurations unchanged except for the AgentAssist URL for new users and the Data Tables for existing users. Copy the “Widget URL” from AgentAssist and paste it into the corresponding fields in Genesys.

Backward Compatibility

  • Existing URLs in Genesys will continue to lead to v2 for existing users.
  • To switch to v3.0, copy the “Widget URL” from AgentAssist and paste it into the Genesys Data Table for existing users.