Playground
The Playground provides a real-time environment to test AI Chat agents, preview responses, and validate tone, behavior, and knowledge before going live.
Introduction
The Playground lets your team interact with your AI Chat agent in real time, simulate customer conversations, and review responses to ensure tone, knowledge, and behavior match your business standards. It is intended for testing and validation only and is not connected to live customer channels.
This can be accessed by clicking it on the left sidebar to open the chat interface and start testing your AI Chat agents.
Why use the Playground
Preview agent behavior. Test how the AI responds across different customer scenarios.
Ensure consistency. Verify that tone, script, and response style match your business standards.
Test before going live. Validate workflows, greetings, and conversation flows without affecting real customers.
Troubleshoot and refine. Identify knowledge gaps, improve responses, and update scripts before deployment.
Selecting an agent to test
At the top-left of the Playground, click the dropdown to choose which AI Chat agent you want to test.
This allows you to switch between different agents without leaving the Playground.
Testing conversations
Send test messages in the Chat with your agent section to evaluate how the agent responds in real time and validate its behavior across different scenarios.
You can use general questions, edge cases, or service-specific inquiries to review responses and identify areas that may need adjustment.
Restarting a chat
To start fresh, click Restart chat in the upper-right corner of the Playground.
This clears the current conversation and resets the context, allowing you to test a new scenario from the beginning.
Playground side panel
On the right side of the Playground, there is an information panel with four tabs: Editor, Agent overview, Ticket info, and Questions.
Editor tab
The Editor tab uses Copilot AI to help improve your agent’s Knowledge base while testing conversations.
If the agent responds with incorrect information or does not know how to answer a question, you can use the Editor to suggest updates directly from the Playground. This can include:
Updating information in the Business overview
Modifying existing Custom questions
Adding new Custom questions
Suggested changes are generated for review before they are applied. The Editor can only update content within the Business overview and Custom questions sections of the Knowledge Base.
A Restart chat button is available below the Editor text box to clear the current Copilot AI conversation and start a new editing session.
Agent overview tab
The Agent overview tab displays configuration details for the active test agent, including name, last updated time, tone, response length, and specialization.
You can click View agent settings to open the Agents page and make adjustments.
Ticket info tab
The Ticket info tab shows simulated contact and ticket data based on the current test conversation.
This section is for preview purposes only. The Playground is not connected to Contacts, Tickets, or Inbox, and it does not create or update real records. Even if you use existing contact details, no data will appear in the Inbox and no ticket will be generated.
If a ticket would normally be created based on the conversation, this tab shows a preview of how it would appear, including contact details and a simulated ticket view with title, type, status, summary, and related conversation data.
Questions tab
The Questions tab provides pre-written prompts that you can click to send into the chat.
These help you quickly test scenarios such as general inquiries, pricing questions, scheduling requests, or custom business-specific cases.
Last updated