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 Events
LLM and Generative AI
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

2.5.x Release Notes

This document provides information on the various releases and the corresponding feature updates and enhancements introduced in version 2.5.x of the AgentAssist application.

V2.5.4 Sep 10, 2023

Patch Release

This update includes bug fixes.

V2.5.3 Aug 19, 2023

This update includes feature enhancements and bug fixes.

FEATURE ENHANCEMENT
AGENT COACHING
None Intent Rule The None Intent rule contains utterances that are used to prevent false positives during intent identification. By using these utterances, you can train the system to differentiate between a positive (trained) and a false-positive (wrong, untrained, or ambiguous) utterance.

This rule is enabled by default, and you can’t disable or delete it. You can, however, add more utterances to make this rule more effective. Learn More

Price Objection Rule This rule contains a customer’s price objection utterance and defines the actions agents should take in response. It is a predefined rule with conditions and actions; however, you can customize them, but can’t edit or delete the rule name. You can disable this rule, if needed. Learn More
Talk Ratio Agent to customer talk ratio – the proportion of speaking time by an agent compared to a customer. These figures help supervisors determine an agent’s engagement and performance in a conversation with a user. By default, it can be identified only once. Learn More
INTEGRATIONS
LLM & Generative AI Integration AgentAssist now provides out-of-box integration with OpenAI and Azure OpenAI models. Bot designers can leverage the model’s advanced capabilities across Generative AI features in AgentAssist. The integration improves AgentAssist’s performance through auto-generated context-specific suggestions. Learn More
Genesys AgentAssist Integration – Auto Summarisation at the end of the conversation AgentAssist automatically generates a summary of the conversation when it ends. Agents can copy the summary and save it into the Notes or Remark field while closing the conversation. Learn More

V2.5.2 August 12, 2023

Patch Release

This update includes feature enhancements and bug fixes.

FEATURE ENHANCEMENT
Event Listener for Genesys For conversation logs, communications from both the user and the agent need to be captured. We were not able to capture agent responses in the Genesys agent desktop. Hence, for Genesys, we have introduced Event Listener to record agent communications.
Enable/Disable AgentAssist inside SmartAssist By default, AgentAssist Widget is enabled inside the SmartAssist Agent Console. You can disable/enable AgentAssist Widget or an individual AgentAssist feature, as required. Learn more
Rule Description Tooltip Rule description is mandatory as LLM uses it to suggest similar utterances. A tooltip is introduced beside the Rule Description header to help users write better descriptions.
Further Improvements–Rules The Rules configuration process has been improved further:

  • A triangular exclamation mark is introduced to indicate incomplete or invalid rule configuration.
  • An information mark, an “i” inside a circle, is introduced beside the Tags header that reads, “Press Enter to add.”
When Entity Value is not Identified For an invalid Entity Value, now an error prompt is displayed with the expected format/value. For example, if the expected entity is a number and the user enters another value than a number, the following error message is displayed – “Expected Format: Number.”
Template Support for Third-Party AgentAssist Third-Party systems do not support a few AgentAssist templates due to specific bot SDKs. In such scenarios, a warning message is displayed that reads, “Templates are not supported, visit your bot builder to disable.”
Change of Groups to Tags To enhance the Agent Coaching module, we have now replaced Groups with Tags. We have also introduced “Channels.” While Tags help in identifying and categorizing rules, Channels determine the communication channels through which rules are applied.
Improvement in Speech Analysis-Based and Utterance-Based Triggers We have improved the process flow of Speech Analysis-based triggers such as Cross Talk, Dead Air, and Speech Speed. Also, we have improved the process flow of Utterance-based triggers.
AgentAssist login using SmartAssist Credentials When users try to log in to AgentAssist using their SmartAssist credentials, they are redirected to the AgentAssist account opening page through the Create New Account button.

 

V2.5.1 July 22, 2023

Patch Release

This update includes feature enhancements and bug fixes.

Introducing Real-time Agent Coaching (Beta)

Contact center agents are often under pressure to provide better customer experiences. Dealing with high call volumes and complex conversations can make it challenging for agents to meet or exceed customer expectations without sufficient support, guidance, and tools.

The new Agent Coaching feature enables real-time coaching based on the analysis of agent utterances and speech. You can set up specific utterances and speech patterns for AgentAssist to take relevant actions, such as providing hints and nudges or emailing the manager when those patterns occur.

As a contact center supervisor, you can view triggered utterances, including the frequency of their occurrence, and evaluate how well agents follow the recommended actions within a specific timeframe. You can also review speech patterns such as cross talks, dead air, and speech speed to help agents optimize their communication style.

Key Agent Coaching features

  • Organize Agent Coaching in Groups (Beta): You can create new agent coaching groups and delete or modify the existing groups. You can also create new rules and modify, delete, disable, and shuffle among groups existing rules.
  • Set Speech-Based Triggers (Beta): Set triggers based on the speech pattern or event of the customer and the agent and provide real-time feedback and coaching to improve customer-agent conversation quality:
    • Dead Air detects the silence/no talk duration and frequency during a conversation. Users can set up the duration in seconds and the frequency for this feature to get triggered. They can also select the agent or the customer as the subject who is silent.
    • Cross-talk detects cross-talk or two parties talking simultaneously during a conversation. Users can set up the cross-talk duration in seconds and its frequency for this feature to be triggered.
    • Speech-Speed detects the unusually fast speech pattern during a conversation. Users can select either the customer or the agent as the subject. Then, they set up the number of words per minute and the frequency for this feature to be triggered.
  • Set up Nudge, Hint, and Email Manager Actions for a Trigger: The action types that you can configure to determine what actions AgentAssist should take for a trigger:
    • Nudge Action (Beta): The agent receives a small pop-up message on their screen that disappears after 5 seconds. You can set up positive, neutral, warning, and critical nudges and enter a message for the agent.
    • Hint Action (Beta): Small pop-up messages that appear on the agent’s screen while communicating with the customers, advising agents to take further action. You can set up positive, neutral, warning, and critical hints and enter a message for the agent.
    • Email Manager Action (Beta): A push notification alert to the manager’s desktop. You can customize and set who should receive this notification. It applies to the customer’s actions, such as using profanity or too many cross-talks. Instead of nudges and hints, the system can send an alert notification to the manager in these situations.
  • Measure Agents’ Adherence to Nudges and Hints: Use the Add Adherence feature to know the agent’s adherence to the instructions in the form of nudges and hints. You can use this acknowledgment to measure agent performance on the Dashboard.

To learn more about the Agent Coaching feature, including groups, rules, associated triggers, and actions, see the detailed Agent Coaching document.

v2.5 July 10, 2023

Patch Release

This update includes bug fixes.

2.5.x Release Notes

This document provides information on the various releases and the corresponding feature updates and enhancements introduced in version 2.5.x of the AgentAssist application.

V2.5.4 Sep 10, 2023

Patch Release

This update includes bug fixes.

V2.5.3 Aug 19, 2023

This update includes feature enhancements and bug fixes.

FEATURE ENHANCEMENT
AGENT COACHING
None Intent Rule The None Intent rule contains utterances that are used to prevent false positives during intent identification. By using these utterances, you can train the system to differentiate between a positive (trained) and a false-positive (wrong, untrained, or ambiguous) utterance.

This rule is enabled by default, and you can’t disable or delete it. You can, however, add more utterances to make this rule more effective. Learn More

Price Objection Rule This rule contains a customer’s price objection utterance and defines the actions agents should take in response. It is a predefined rule with conditions and actions; however, you can customize them, but can’t edit or delete the rule name. You can disable this rule, if needed. Learn More
Talk Ratio Agent to customer talk ratio – the proportion of speaking time by an agent compared to a customer. These figures help supervisors determine an agent’s engagement and performance in a conversation with a user. By default, it can be identified only once. Learn More
INTEGRATIONS
LLM & Generative AI Integration AgentAssist now provides out-of-box integration with OpenAI and Azure OpenAI models. Bot designers can leverage the model’s advanced capabilities across Generative AI features in AgentAssist. The integration improves AgentAssist’s performance through auto-generated context-specific suggestions. Learn More
Genesys AgentAssist Integration – Auto Summarisation at the end of the conversation AgentAssist automatically generates a summary of the conversation when it ends. Agents can copy the summary and save it into the Notes or Remark field while closing the conversation. Learn More

V2.5.2 August 12, 2023

Patch Release

This update includes feature enhancements and bug fixes.

FEATURE ENHANCEMENT
Event Listener for Genesys For conversation logs, communications from both the user and the agent need to be captured. We were not able to capture agent responses in the Genesys agent desktop. Hence, for Genesys, we have introduced Event Listener to record agent communications.
Enable/Disable AgentAssist inside SmartAssist By default, AgentAssist Widget is enabled inside the SmartAssist Agent Console. You can disable/enable AgentAssist Widget or an individual AgentAssist feature, as required. Learn more
Rule Description Tooltip Rule description is mandatory as LLM uses it to suggest similar utterances. A tooltip is introduced beside the Rule Description header to help users write better descriptions.
Further Improvements–Rules The Rules configuration process has been improved further:

  • A triangular exclamation mark is introduced to indicate incomplete or invalid rule configuration.
  • An information mark, an “i” inside a circle, is introduced beside the Tags header that reads, “Press Enter to add.”
When Entity Value is not Identified For an invalid Entity Value, now an error prompt is displayed with the expected format/value. For example, if the expected entity is a number and the user enters another value than a number, the following error message is displayed – “Expected Format: Number.”
Template Support for Third-Party AgentAssist Third-Party systems do not support a few AgentAssist templates due to specific bot SDKs. In such scenarios, a warning message is displayed that reads, “Templates are not supported, visit your bot builder to disable.”
Change of Groups to Tags To enhance the Agent Coaching module, we have now replaced Groups with Tags. We have also introduced “Channels.” While Tags help in identifying and categorizing rules, Channels determine the communication channels through which rules are applied.
Improvement in Speech Analysis-Based and Utterance-Based Triggers We have improved the process flow of Speech Analysis-based triggers such as Cross Talk, Dead Air, and Speech Speed. Also, we have improved the process flow of Utterance-based triggers.
AgentAssist login using SmartAssist Credentials When users try to log in to AgentAssist using their SmartAssist credentials, they are redirected to the AgentAssist account opening page through the Create New Account button.

 

V2.5.1 July 22, 2023

Patch Release

This update includes feature enhancements and bug fixes.

Introducing Real-time Agent Coaching (Beta)

Contact center agents are often under pressure to provide better customer experiences. Dealing with high call volumes and complex conversations can make it challenging for agents to meet or exceed customer expectations without sufficient support, guidance, and tools.

The new Agent Coaching feature enables real-time coaching based on the analysis of agent utterances and speech. You can set up specific utterances and speech patterns for AgentAssist to take relevant actions, such as providing hints and nudges or emailing the manager when those patterns occur.

As a contact center supervisor, you can view triggered utterances, including the frequency of their occurrence, and evaluate how well agents follow the recommended actions within a specific timeframe. You can also review speech patterns such as cross talks, dead air, and speech speed to help agents optimize their communication style.

Key Agent Coaching features

  • Organize Agent Coaching in Groups (Beta): You can create new agent coaching groups and delete or modify the existing groups. You can also create new rules and modify, delete, disable, and shuffle among groups existing rules.
  • Set Speech-Based Triggers (Beta): Set triggers based on the speech pattern or event of the customer and the agent and provide real-time feedback and coaching to improve customer-agent conversation quality:
    • Dead Air detects the silence/no talk duration and frequency during a conversation. Users can set up the duration in seconds and the frequency for this feature to get triggered. They can also select the agent or the customer as the subject who is silent.
    • Cross-talk detects cross-talk or two parties talking simultaneously during a conversation. Users can set up the cross-talk duration in seconds and its frequency for this feature to be triggered.
    • Speech-Speed detects the unusually fast speech pattern during a conversation. Users can select either the customer or the agent as the subject. Then, they set up the number of words per minute and the frequency for this feature to be triggered.
  • Set up Nudge, Hint, and Email Manager Actions for a Trigger: The action types that you can configure to determine what actions AgentAssist should take for a trigger:
    • Nudge Action (Beta): The agent receives a small pop-up message on their screen that disappears after 5 seconds. You can set up positive, neutral, warning, and critical nudges and enter a message for the agent.
    • Hint Action (Beta): Small pop-up messages that appear on the agent’s screen while communicating with the customers, advising agents to take further action. You can set up positive, neutral, warning, and critical hints and enter a message for the agent.
    • Email Manager Action (Beta): A push notification alert to the manager’s desktop. You can customize and set who should receive this notification. It applies to the customer’s actions, such as using profanity or too many cross-talks. Instead of nudges and hints, the system can send an alert notification to the manager in these situations.
  • Measure Agents’ Adherence to Nudges and Hints: Use the Add Adherence feature to know the agent’s adherence to the instructions in the form of nudges and hints. You can use this acknowledgment to measure agent performance on the Dashboard.

To learn more about the Agent Coaching feature, including groups, rules, associated triggers, and actions, see the detailed Agent Coaching document.

v2.5 July 10, 2023

Patch Release

This update includes bug fixes.