- Home
- /
- Article
Thanks for your feedback.
Cisco AI Assistant for Webex Contact Center: Administrator's guide to configuring suggested responses
In this article
Feedback?Disclaimer: This comprehensive guide consolidates individual help articles into a single, navigable document, providing a structured learning path for your role. As this guide is compiled from standalone articles, you may encounter 'What's next' links within certain sections. Please note that for content directly related to the current topic, the information referenced by these links is typically presented as the subsequent chapter within this guide. Links that direct you to information about other Webex Contact Center features or broader Webex topics will lead you to relevant articles on the Webex Help Center.
Deployment workflow for suggested responses
This article provides administrators with a workflow for deploying and optimizing the suggested responses feature in Webex Contact Center. Suggested responses empower agents with real-time, AI-driven guidance to enhance customer interactions and operational efficiency. This feature leverages AI Assistant skills, defined and managed in Webex AI Studio, to analyze conversation transcripts and offer contextually relevant responses or actions to agents.
As an administrator, you are responsible for setting up, configuring, and maintaining this feature. This article serves as a central navigation point, outlining the end-to-end deployment and management process and linking to detailed articles for each step. Following this workflow ensures a successful and optimized implementation of the feature in your organization.
Implementing suggested responses involves a logical sequence of steps, from initial activation to ongoing optimization. The process is divided into four primary phases:
Step 1: Activating the feature — Enablement
Begin by activating the suggested responses feature in Control Hub and ensuring all prerequisites are met. This foundational step is required for further configuration, and involves the following activities:
- Enable suggested responses in Control Hub.
- Verify licensing requirements.
- Ensure supporting services, such as real-time transcriptions are ready.
Reference article: Enable suggested responses for your Webex Contact Center
Step 2: Defining AI intelligence — Building AI Assistant skills
Create the AI intelligence that powers suggested responses using the Webex AI Studio. This step involves the following activities:
-
Create new AI Assistant skills in Webex AI Studio.
-
Define the goals for each skill.
-
Link skills to relevant Knowledge Bases.
-
Configure specific actions for each skill.
Reference articles:
Step 3: Deployment — Connecting intelligence to operations
Integrate configured AI Assistant skills into your live contact center environment by assigning them to specific queues. This step involves the following activity:
-
Assign AI Assistant skills to queues.
Reference article: Link AI Assistant skills to queues
Step 4: Monitoring and optimization — Ensuring performance and continuous improvement
After deployment, verify functionality, monitor performance, and iteratively refine your configurations for maximum effectiveness. This step involves the following activities:
-
Test AI Assistant responses using preview features.
-
Analyze impact on agent performance.
-
Use session data for auditing and debugging.
Reference article: Test and monitor suggested responses performance
Deployment is an iterative process. Regularly monitor the performance of suggested responses and use insights from monitoring and optimization to adjust your AI Assistant skills and actions. This continuous improvement loop helps ensure the feature remains aligned with evolving customer needs and operational requirements.
Enable suggested responses for your Webex Contact Center
This article explains how Contact Center administrators can enable and manage the Suggested Responses feature for agents, helping improve communication efficiency, agent performance, and customer satisfaction.
Suggested Responses is a real-time AI Assistant feature in Webex Contact Center that enhances agent productivity and customer satisfaction. This feature leverages advanced artificial intelligence (AI) to provide contextual guidance to human agents during both inbound and outbound voice interactions and inbound digital customer interactions, supporting both voice and digital channels. Suggested Responses aims to transform the agent experience by offering timely, relevant suggestions on what to say and what actions to take directly within the Agent Desktop. This proactive assistance streamlines workflows, reduces response times, and ensures consistent, high-quality service delivery.
As a Webex Contact Center administrator, you can enable and manage the suggested responses feature for your organization. This process involves ensuring your system meets the necessary prerequisites and activating the feature within Control Hub. Once enabled, you can configure AI Assistant skills to tailor the suggestions provided to your agents.
Who can use this feature
Suggested responses is useful for the following stakeholders:
-
Agents: Benefit from real-time contextual suggestions, reducing the effort required to find information and improving response accuracy.
-
Administrators: Manage and configure suggested responses settings at the organization and queue levels, create and optimize AI Assistant skills, test and debug configurations, monitor performance, and continuously optimize the feature for their contact center.
Benefits
Implementing suggested responses delivers significant benefits across your contact center:
-
Enhanced agent efficiency: Agents spend less time searching for information or consulting colleagues, leading to quicker resolutions and improved productivity.
-
Reduced handle times: Immediate, accurate suggestions help agents resolve queries faster, directly impacting Average Handle Time (AHT).
-
Consistent customer experience: Ensures all agents, regardless of experience, provide accurate and consistent information, enhancing the overall customer experience.
-
Reduced after-call work: Agents can complete necessary actions and documentation during the interaction, minimizing post-call tasks.
-
Improved first contact resolution: With immediate access to relevant information and actions, agents are better equipped to resolve customer issues on the first contact.
-
Reduce onboarding times for new hires: New agents gain confidence and proficiency faster with real-time AI guidance.
-
Scalability: Supports your growing contact center by enabling agents to handle higher volumes of interactions more effectively.
-
Dual-channel capability: Dual-channel capability: Supports both inbound and outbound voice calls, as well as inbound digital interactions, providing real-time suggestions whether agents engage with customers through phone calls or digital channels like chat or email.
How suggested responses work
Suggested responses integrates seamlessly into your contact center operations through the following high-level flow:
-
Interaction begins: A customer initiates a voice or digital interaction that is routed to an agent through a configured queue, or an agent initiates an outbound voice interaction.
-
Real-time transcription: For voice interactions, the conversation is transcribed in real-time. For digital interactions, text is captured. For more information, see the Enable real-time transcripts for agents article.
-
Agent receives suggestions: The agent can click Get Suggestions button in their desktop, or suggestions may proactively appear based on the conversation context.
-
AI Assistant skill processes: The AI Assistant skill, linked to the specific queue and powered by its knowledge base and defined actions, analyzes the real-time conversation (transcript for voice, text for digital).
-
Suggestions generated: The AI Assistant skill generates relevant text suggestions for replies or proposes actions based on the customer's query.
-
Agent reviews and acts: The agent reviews the suggestions and chooses to use them as-is, modify them, or execute the suggested actions.
-
Continuous support: Suggestions continue to be provided throughout the interaction, adapting to the evolving conversation.
Usage guidelines
The suggested responses feature requires careful management to ensure ethical use and enhance communication. As an administrator, you are responsible for informing agents about the feature's enablement and its AI-powered nature. Ensure UI responsiveness is tested across different devices for consistent functionality.
For more detailed information on usage guidelines and privacy, refer to the Agent Answers and Suggested Responses - AI Transparency Note.
Prerequisites
Before enabling suggested responses, ensure that your Webex Contact Center organization meets the following requirements:
-
Webex Contact Center Flex 3.0: Your contact center must be running on the Flex 3.0 platform.
-
AI Assistant add-on SKU: Your organization must have purchased the A-FLEX-AI-ASST add-on SKU. This entitlement grants access to all AI Assistant features, including suggested responses. For information on managing your subscription and usage, refer to the Usage and Billing Documentation.
-
Media forking for voice interactions: For voice interactions, media forking must be enabled in Flow Designer for the relevant queues. See the Enabling media streaming for specific queues article.
-
Webex Connect enabled: If you plan to configure and use AI-suggested actions that involve external system integrations and fulfillment flows, Webex Connect must be enabled and configured for your organization.
-
Required user permissions: You must have the necessary Role-Based Access Control (RBAC) privileges to access and modify contact center settings in the Control Hub. Specifically, your role should include access to the AI Assistant - Suggested Responses feature. For more information, see the Manage user profiles in Webex Contact Center article.
How to enable suggested responses
Follow these steps to activate the suggested responses feature for your Webex Contact Center:
| 1 |
Sign in to Control Hub and navigate to Services > Contact Center. |
| 2 |
Go to AI Assistant under Desktop Experience in the navigation pane. |
| 3 |
Toggle on Suggested responses to enable the feature at the organization level. |
| 4 |
Click Assign AI Assistant skills and select the AI skill you want to assign to your queue. Add one or more queues to the selected skill and save your changes. These skills determine the type of suggestions agents will receive in each queue. If the specific AI Assistant skill you wish to assign is not yet listed, or for a first-time setup where no skills have been created, click Manage AI Assistant skills to create a new skill in the Webex AI Studio. After creating the skill, return to this screen and complete the assignment process. For more information on assigning skills to queues, refer to the Link AI Assistant skills to queues article. |
Create and manage AI Assistant skills
AI Assistant skills are core components that power the suggested responses feature in Webex Contact Center. These intelligent configurations define how the AI Assistant provide real-time guidance to human agents. As an administrator, you use the Webex AI Studio to create, configure, and manage these skills, tailoring them to the specific needs of your contact center’s queues and interaction types.
This article guides you through accessing the Webex AI Studio, understanding its environment, and the detailed process of creating and managing AI Assistant skills.
Accessing Webex AI Studio
To access the Webex AI Studio, follow the instructions:
| 1 |
Sign in to Control Hub and navigate to Services > Contact Center. |
| 2 |
Go to AI Assistant under Desktop Experience in the navigation pane. |
| 3 |
Under the Suggested responses section, click the Manage AI Assistant skills link. The AI Assistant skills dashboard opens in a new browser tab within the Webex AI Studio. |
Understanding Webex AI Studio
The Webex AI Studio provides dedicated dashboards for managing AI Assistant skills and AI Agents.
- If your organization uses only suggested responses, the AI Assistant skills dashboard is displayed by default.
- If your organization also uses AI Agents, both the AI Assistant skills and AI Agents dashboards are available, allowing you to switch between them, as needed.
Understanding AI Assistant skills Dashboard
The AI Assistant skills dashboard serves as your central hub for comprehensive management of all AI Assistant skills configured for suggested responses. It provides an overview of your existing skills and quick access to various management functions.
Dashboard display
On the dashboard, AI Assistant skills are presented as cards or rows in a list view. Each entry provides key details for quick reference:
- Skill Name: The descriptive name you assigned to the AI Assistant skill.
- Queues: The number of queues the skill is currently linked to.
- Assistant's Goal: A brief description of the skill's purpose.
- Last Modified: The date and time the skill was last updated.
- Last Modified By: The user who last modified the skill.
- Creation Date: The date the skill was created.
- Created By: The user who created the skill.
Dashboard actions
From the main dashboard, you can perform the following actions to manage your AI Assistant skills efficiently:
- Create a new AI Assistant skill: Click + Create skills to begin configuring a new AI Assistant skill from scratch.
- Import AI Assistant skill: Use Import skills to upload AI Assistant skills in JSON format. This allows you to quickly add pre-configured skills or restore backups from external sources.
- Search and filter: Utilize the search bar to quickly find skills by name, linked queue, or last modified date.
- Preview: Open the skill's preview widget to test its responses in a simulated environment.
Individual skill management
For each individual AI Assistant skill listed on the dashboard, you can access additional management options:
- Modify skill settings: Click on any skill's name or card to open its configuration page and modify its settings, including its profile, knowledge base, and actions.
-
Quick actions: Access quick actions for a specific skill by clicking the ellipsis
icon (three dots) to perform tasks such as:
- Pin: Pin the skill to the top of your dashboard for quick access.
- Copy Access Token: Copy the skill's access token for seamless integration, authentication, and testing in external applications or development environments.
- Export: Export the skill's details (in JSON format) to your local folder for backup or transfer.
- Delete: Permanently delete the AI Assistant skill.
If you attempt to delete a skill that is linked to one or more queues, a confirmation message appears, informing you that the skill is in use. You must confirm before proceeding. If confirmed, the Control Hub is notified of the deletion.
Creating AI Assistant skills
Follow these steps to configure an AI Assistant skill that provides relevant suggestions to your agents:
| 1 |
On the AI Assistant skills dashboard, click + Create skills. |
| 2 |
Select Start from scratch and click Next. |
| 3 |
Enter a clear, descriptive name in the Skill name field. The System ID field, which uniquely identifies the skill, is automatically populated based on the Skill name. |
| 4 |
Provide a concise goal in the Goal field. Example:“You are a helpful and polite assistant that helps agents answer queries regarding lost baggage and recommend necessary actions.” For best practices, refer to the Defining goal section. |
| 5 |
Click Create. |
| 6 |
On the Skill Configuration screen, the Profile tab is selected by default. In the Instructions text box, provide detailed, step-by-step guidance for the AI’s behavior and response generation. For best practices, refer to the Crafting instructions section.
|
| 7 |
Switch to the Knowledge tab to select a relevant knowledge base. The skill uses this source to generate suggestions. If a suitable knowledge base does not exist, click + Add new. After creating it, you will return to the skill configuration panel.
For best practices on organizing a knowledge base, refer to the Structuring Knowledge bases section. |
| 8 |
Go to the Actions tab to enable, disable, edit, or delete actions. You can also create new actions by clicking + New action. Actions enable AI skills to suggest or perform specific tasks (such as creating cases, retrieving information, sending emails, or performing integrations), in addition to suggesting information that can be used to respond to the customers. If you are not connecting to a knowledge base, add at least one action. For best practices, refer to the Defining actions section. For detailed instructions, refer to the Configuring actions for AI Assistant skills article. |
| 9 |
When all required fields are complete, click Save changes. Before publishing a skill, test its behavior using the Preview button. This allows you to validate the relevance and accuracy of suggestions. For detailed information, refer to the Previewing AI Assistant skill responses section. |
| 10 |
Click Publish to finalize the skill. |
Note on AI engine: The AI model used for this feature is GPT-4o-mini, to ensure optimal performance and reliability.
What's next?
After creating and configuring your AI Assistant skills, proceed with the next steps in deploying suggested responses for your contact center:
- Assign AI Assistant skills to queues: To make your skills available to agents, assign them to the relevant contact center queues. Refer to the Link AI Assistant skills to queues article.
Once your skills are assigned and actions are configured (if needed), continue to test and monitor the suggested responses feature to ensure optimal agent performance and customer experience. For guidance on monitoring and continuous improvement, refer to the Test and monitor suggested responses performance article.
Link AI Assistant skills to queues
To ensure that AI Assistant's suggested responses provide relevant and timely guidance, you must accurately link AI Assistant skills to the appropriate queues in your Webex Contact Center. This configuration ensures that when a human agent handles an inbound or outbound customer interaction from a specific queue, the system knows exactly which AI skill should provide suggestions based on the type of customer query handled by that queue.
This article provides instructions for linking AI Assistant skills to queues, along with important considerations for effective deployment and optimal performance.
Benefits
Linking AI skills to queues offers several advantages for contact center operations:
-
Targeted suggestions: Targeted suggestions: Agents receive contextually relevant suggestions tailored to the specific type of inbound or outbound interactions handled by each queue.
-
Optimized performance: AI processing resources are used efficiently by activating the correct AI skill only when interactions flow through its associated queues, minimizing unnecessary processing.
-
Scalability: Enables management of a diverse range of customer inquiries across multiple queues, each supported by a specialized AI skill, supporting organizational growth and adaptation.
-
Enhanced agent efficiency: Context-aware suggestions help agents respond more quickly and accurately, reducing handle times and improving productivity.
Operational notes and limitations
Follow these notes and considerations for effective linking of AI skills to queues:
-
One queue, one AI skill: A single queue can be linked to one AI skill at a time to prevent conflicting suggestions for the same interaction.
-
One AI skill, multiple queues: A single AI skill can be linked to multiple queues, which is useful if several queues handle similar types of inquiries.
-
Media streaming for voice channels: For voice interactions, media streaming must be enabled for all queues linked to AI skills. Media streaming duplicates the audio stream, enabling real-time transcription which is essential for the AI skill to generate suggestions. For details, see Enable real-time transcripts for agents.
-
Impact of removing queues: If you attempt to unassign a queue from an AI skill, a confirmation message will appear, especially if active calls may be affected.
-
Batch unassignment: If an AI skill is linked to more than 50 queues, you may need to unassign them in batches due to system limitations.
-
Deletion of AI skills: An AI skill cannot be deleted while it is still linked to any queues. Unassign all queues before deleting a skill.
Prerequisites
Before linking AI skills to queues, ensure the following requirements are met:
-
Suggested responses enabled: Ensure that the Suggested responses feature toggle is enabled. For instructions, refer to the Enable suggested responses for Webex Contact Center.
-
AI Assistant add-on SKU: Your organization must have purchased the AI Assistant add-on SKU (A-FLEX-AI-ASST) for Webex Contact Center.
-
Webex AI Studio access: You must have administrator privileges to access the Webex AI Studio platform.
-
Configured AI skill: At least one AI skill must be created, configured, and published in Webex AI Studio. For more information, see Create and manage AI Assistant skills.
-
Configured queues: The target queues must be set up in your Webex Contact Center. For more information, refer to the Create queues and configure routing patterns article.
Linking AI skills to queues
To ensure agents receive precise, context-aware suggestions, link your AI skills to the appropriate queues as follows:
| 1 |
Sign in to Control Hub and navigate to Services > Contact Center. |
| 2 |
In the navigation pane, go to AI Assistant under Desktop Experience. |
| 3 |
Under Suggested responses, click Assign AI Assistant skills to select the AI skills and the queues you want to link. |
| 4 |
Select an AI skill and locate the Assigned Queues section, which displays the queues currently linked to this skill. |
| 5 |
Click Add Queues. |
| 6 |
In the Select Queues screen, use the search bar to find specific queues by name. |
| 7 |
Select the check boxes next to the queues you want to link to this AI skill. You can select up to 50 queues at a time. To link additional queues, repeat this process. |
| 8 |
Click Save to assign the selected queues to the AI skill. |
Test and monitor suggested responses performance
After configuring suggested responses and setting up your AI Assistant skills, it is crucial to test their effectiveness and continuously monitor their performance. This proactive approach ensures that the AI Assistant provides accurate and relevant suggestions, leading to improved agent efficiency and enhanced customer satisfaction. Webex Contact Center provides comprehensive tools within Webex AI Studio and Analyzer for thorough testing and ongoing monitoring.
This article guides you through the processes of previewing AI Assistant skill responses, monitoring their impact using performance metrics, and utilizing Sessions and History tabs for auditing and debugging.
Prerequisites
Before testing and monitoring suggested responses performance, ensure the following:
- AI Assistant add-on SKU: Your organization must have purchased the AI Assistant add-on SKU (A-FLEX-AI-ASST) for Webex Contact Center.
- Webex AI Studio access: You must have the necessary administrator privileges to access the Webex AI Studio platform.
-
Configured AI Assistant skill: An AI Assistant skill must be created, configured,
and published in the Webex AI Studio with relevant knowledge bases, instructions, and
actions.
For more information, refer to Create and manage AI Assistant skills and Configure actions for AI Assistant skills articles.
-
Linked AI Assistant skill to queues: The AI Assistant skill providing suggested
responses must be linked to the appropriate queues.
For more information, refer to the Link AI Assistant skills to queues article.
-
Real-time transcriptions enabled (for voice): For voice interactions, real-time
transcription must be enabled for the relevant queues. This is essential for the AI
Assistant to process spoken conversations and generate suggestions.
For more information, refer to the Enable real-time transcripts for agents article.
Previewing AI Assistant skill
Before deploying an AI Assistant skill to live agents, you can test its behavior in a simulated environment within the Webex AI Studio. This allows you to validate the relevance and accuracy of suggestions. Follow the steps below:
| 1 |
Sign in to Control Hub and navigate to Services > Contact Center. |
| 2 |
Go to AI Assistant under Desktop Experience in the navigation pane. |
| 3 |
Under the Suggested responses section, click the Manage AI Assistant skills link. The AI Assistant skills dashboard opens in a new browser tab within the Webex AI Studio. |
| 4 |
In the AI Assistant skills dashboard, click on the specific AI Assistant skill you want to test. This will open its configuration page. |
| 5 |
Click the Preview button. |
Chat mode preview: The preview within the Webex AI Studio allows you to simulate a chat interaction. You can assume the role of a customer, type queries, and observe how the AI Assistant skill generates suggestions, just as a human agent would see them.
Note for voice channel testing: For a detailed preview of how suggestions appear and function during live voice interactions, you must switch to the Agent Desktop and test the feature in a real-call scenario. This requires real-time transcription to be active for the call.
Utilizing Analyzer for performance monitoring
The suggested responses feature is part of the AI Assistant. You can review AI Assistant reports and KPIs in Analyzer to monitor usage and effectiveness.
For more information, refer to the AI Assistant reports in Analyzer article.
Insights offered by suggested responses report
This report provides insights into the real-world impact of suggested responses on agent performance and customer satisfaction, helping you gather feedback for continuous improvement and measure the feature's effectiveness. The report offers insights into the following key areas:
- Usage metrics: Track how often agents use the Get Suggestions button, the number of suggestions offered per interaction, and the feature's adoption rate.
- Impact on KPIs: Monitor changes in Average Handle Time (AHT), consults/transfers, customer satisfaction (CSAT)/ auto CSAT, and First Contact Resolution (FCR) for interactions where suggested responses were used.
- Action execution: Track the number of unmoderated and moderated actions performed and their success rates.
- Real-time transcription KPIs: For voice interactions, cross-reference the performance of suggested responses with the real-time transcription KPIs report to ensure the underlying transcription service is performing optimally. For more information, refer to the Enable real-time transcripts for agents article.
Utilizing AI Studio for auditing and debugging
Besides the Configuration tab, the Webex AI Studio provides two tabs for auditing and debugging your AI Assistant skills: Sessions and History.
Sessions tab
The Sessions tab provides a detailed record of every interaction where an AI Assistant skill was used for suggested responses. It can be used for auditing, debugging, and continuous improvement of AI Assistant skill performance in live interactions. To effectively leverage the insights from this tab, perform the following actions:
-
View and filter interactions: The Sessions page displays a list of all
interactions where AI Assistant skills provided suggestions:
-
- Search: Use the search bar to find specific interactions by contact session ID, consumer ID or interaction ID.
- Filter: Apply filters to narrow down the list by:
- Contact date: Interactions within a specific time range.
- Agents, teams, queues: Interactions handled by specific personnel or routed through particular queues.
- Channel types: Voice or digital interactions.
- Actions suggested/performed: Interactions where specific actions were proposed or performed.
- Errors occurred: To filter sessions in which an error occurred.
- Hide test sessions: To exclude test sessions from your view.
- Agent handover happened: To filter sessions where an agent handover occurred.
- Downvoted: To filter sessions that were downvoted by agents.
- Flagged interactions (if implemented): Interactions flagged by human agents for review.
-
-
Detailed interaction view: Click on any interaction in the list to view
comprehensive details:
- Interaction transcript: The full conversation between the human agent and the customer, provided by the real-time transcriptions feature.
- AI Assistant skill used: Identifies which AI Assistant skill provided suggestions for this interaction.
- List of suggestions: Shows all suggestions provided to the human agent, linked to the specific customer query that prompted them.
- Source of suggestions: Displays the top articles, FAQs, or sections from the knowledge base that were used to generate the suggestions. This allows you to verify the accuracy and relevance of the information.
- Actions suggested and performed: Provides details on which actions were proposed and whether they were performed (in unmoderated or moderated mode).
- Agent modifications: If a human agent edited any fields in a moderated action before submission, these modifications are recorded.
- Additional context: Any extra information provided by the human agent to refine suggestions is visible.
- Metadata: Includes contact session ID, human agent ID/name, team ID/name, queue ID/name, contact date/time, and channel type.
History tab
The History tab provides a record of the configuration changes and versions of your AI Assistant skills. While not directly a performance monitoring tool for live interactions, it is crucial for understanding why performance might have changed due to configuration updates.
- Version history: Tracks different published versions of your AI Assistant skill, allowing you to revert to previous configurations, if needed.
- Change logs: Provides a detailed log of modifications made to the AI Assistant skill's settings, including who made the change, when, and what was changed. This is essential for auditing and debugging configuration-related issues.
Strategies for deploying AI Assistant skills
Best practices for optimizing AI Assistant skills
AI Assistant skills are the intelligent core of the suggested responses feature in Webex Contact Center. Unlike AI Agents that interact directly with customers, AI Assistant skills are designed to empower and guide your human agents in real time. Optimizing these skills involves careful crafting of their goals, instructions, knowledge base structure, and action definitions to ensure accurate, relevant, and actionable suggestions.
This article provides best practices for administrators to create highly effective AI skills, focusing on strategic design and content quality. For detailed steps on creating and managing AI skills, defining actions, and linking them to queues, refer to the respective administrator articles.
Best practices
Defining goal
The goal defines the purpose of your AI skill. It is a high-level statement that guides the AI's behavior and clarifies its role in assisting the human agent.
- Focus on agent assistance: Clearly articulate how the AI will help the human agent. The goal should always reflect the AI's role as a support tool for the agent, not as a direct customer-facing entity.
Example: "You are a helpful and polite assistant that will help agents answer queries regarding lost baggage and recommend necessary actions."
- Keep it concise and action-oriented: A short, clear goal helps the AI maintain focus.
- Align with skill capabilities: Ensure the goal is realistic and achievable based on the Knowledge Base content and actions defined for the skill.
For detailed steps on setting the goal, refer to the Create and manage AI Assistant skills article.
Crafting instructions
Instructions provide detailed guidance to your AI skill on how to process information and generate suggestions. This section differentiates your AI Assistant skill from an AI Agent, as these instructions are for the AI to assist the agent.
- Define skill’s persona (as an assistant to the agent): Clearly state the skill's role and expertise as an assistant to the human agent.
Example: "You are an expert AI assistant for agents handling billing inquiries. Your role is to analyze customer conversations and provide agents with the most relevant information and actions to resolve billing questions."
- Outline tasks and decision flow: Break down the overall task into specific, sequential steps from the AI's perspective. Guide the AI on what to look for in the conversation and what type of suggestion or action to provide.
Example: "First, listen for the customer's primary issue regarding their lost baggage. Then, if flight details are needed, suggest the agent use the 'Retrieve Flight Details' action. If a claim needs to be filed, suggest the agent use the 'Create Lost Baggage Claim' action."
- Specify suggestion formatting: Instruct the AI on presenting information clearly and concisely for the agent.
Example: "Suggestions should be presented in clear, bulleted points for easy readability. Action suggestions should clearly state the action name and its purpose for the agent."
- Reference actions clearly: When the AI should suggest an action, explicitly reference the configured action name.
Example: "If the customer mentions 'fraudulent charge,' suggest the agent use the <Initiate Fraud Investigation> action."
- Plan for error handling and fallbacks: Instruct the AI on how to respond when it cannot provide a confident or relevant suggestion.
Example: "If you cannot provide a confident suggestion for the agent's current context, inform the agent by stating: 'No relevant suggestion available. Please refer to the knowledge base or consult a supervisor.'"
- Add guardrails (scope of assistance): Remind the AI to stay within its defined scope.
Example: "Your suggestions must always be focused on assisting the agent with the customer interaction. Do not attempt to answer non-billing questions or interact directly with the customer."
For detailed steps on providing instructions, refer to the Create and manage AI Assistant skills article.
Structuring Knowledge Bases
The Knowledge Base is the factual foundation for your AI skill. Its organization should prioritize information that helps an agent respond or act effectively.
- Agent-centric content: Prioritize information agents frequently need to explain, troubleshoot, or act upon. Focus on content that is directly useful for an agent's workflow.
Example: “For a "Refund Policy" article, include not just the policy text but also common customer questions and practical steps for processing a refund.”
- Organize content logically: Use categories within your KB to group related information. This helps both the AI and agents navigate efficiently and improves suggestion relevance.
- Ensure accuracy and consistency: Verify that all information is accurate and up to date. Avoid conflicting or outdated content.
- Optimize document quality:
- Clarity: Use plain language.
- Conciseness: Be direct; AI and agents need quick answers.
- Structure: Use headings, subheadings, bullet points, and numbered lists for readability and to help the AI extract key information.
- File size: Consider breaking down very large documents into smaller, topic-specific ones to improve retrieval speed and relevance.
- Regular review and update: Continuously review your KB content to ensure it remains relevant and accurate. Update information whenever policies, products, or processes change.
For detailed steps on linking a knowledge base to a skill, refer to the Create and manage AI Assistant skills article.
Defining actions
Actions define specific tasks the AI skill can suggest or perform. When configuring actions, consider their presentation to the agent and their impact on workflow.
- Clearly define action objectives: The Action name and Action description should be clear, concise, and immediately understandable to the agent.
Example: Action name: "Retrieve Order Status"; Action description: "This action will fetch the current status of a customer's order using their order ID."
- Minimize complexity: Keep individual actions simple and focused. Break complex, multi-step processes into smaller, distinct actions.
- Accurately describe user inputs: For each user input (slot), provide a clear description to help the AI accurately identify and collect needed information.
- Choose appropriate fulfillment mode: Educate agents on the differences between
the fulfillment modes:
- Unmoderated mode: For routine, low-risk actions requiring no agent review (example, logging a simple interaction detail). Ensure robust error handling.
- Moderated mode: For actions requiring agent verification, input, or approval (example, submitting a form, confirming sensitive data). This empowers the agent and ensures accuracy.
For detailed information, agents can be referred to the Understand and manage AI-suggested actions article.
- Design for agent workflow: Consider how the action will appear in the Agent Desktop. Use the Card Layout configuration to influence information presentation.
For detailed steps on configuring actions, refer to the Configure actions for AI Assistant skills article.
Testing and iteration
Creating and optimizing effective AI skills is an iterative process. Continuous testing and refinement are essential to ensure ongoing accuracy and relevance.
- Test and preview regularly: Use the preview feature in AI Studio to simulate interactions and verify that your skill generates accurate and relevant suggestions and actions.
- Monitor performance data: Use Analyzer for performance metrics and Session History in AI Studio for detailed auditing and debugging of interactions. This data is crucial for identifying areas for refinement.
- Refine based on feedback: Pay close attention to agent feedback on suggestions and actions. Use this feedback, along with insights from session history analysis, to refine your skill's goal, instructions, Knowledge base content, and action definitions. This ensures your AI skill remains effective and adapts to evolving agent needs and contact center operations.
For detailed steps on testing and monitoring, refer to the Test and monitor suggested responses performance article.
AI terminology and concepts in Webex Contact Center
To effectively utilize the AI capabilities in Webex Contact Center, it is important to understand key terminology and how various AI components interact. This article provides a glossary of essential AI concepts, helping you navigate and leverage intelligent features designed to enhance customer interactions and operational efficiency.
AI terminology
-
AI Assistant: The AI capability within Webex Contact Center designed to enhance agent performance. Features such as real-time transcripts, suggested responses, AI-generated summaries, automated wellness breaks, auto CSAT, and topic analytics are core components of the AI Assistant.
-
AI Assistant skills: Configurable entities created and managed in AI Studio that power the suggested responses feature. These skills assist agents in real time by providing contextual suggestions and actions, serving as intelligent guides.
-
AI Agents: Distinct from AI Assistant skills, AI Agents independently handle customer interactions (voice or digital) without immediate human intervention. They can answer Frequently Asked Questions (FAQs), process routine requests, or route customers, often serving as the first point of contact before escalating to a human agent. For more information, refer to the Webex AI Agent.
-
AI Studio: The central platform in Webex Contact Center where administrators create, manage, and configure both AI Assistant skills (for agent assistance) and AI Agents (for autonomous interactions). Depending on your organization's enabled features, you may see one or both dashboards within the AI Studio.
-
Knowledge Base (KB): A centralized repository of information (such as FAQs, articles, and documents) that AI Assistant skills and AI Agents use to generate accurate and contextually relevant responses.
-
Actions: Predefined tasks or workflows that an AI Assistant skill can suggest to an agent or execute on the agent's behalf (with or without agent's review). For AI Agents, actions are tasks they perform autonomously.
-
Fulfillment modes: Define how an AI Assistant skill executes actions:
-
Moderated mode: The AI Assistant skill collects information for an action and presents it to the agent for review and explicit approval before execution.
-
Unmoderated mode: The action is executed independently by the AI Assistant skill, without requiring agent intervention or approval.
-
