Playground

What is the Playground?

The Playground is a testing environment in the Advanced Editor where you can interact with your agent in real-time. It allows you to simulate conversations, test different scenarios, and validate your agent’s behavior before deploying to production.

Testing in the Playground saves time and money by catching issues before they affect real customer calls.

Accessing the Playground

The Playground is located in the right panel of the Advanced Editor. Click the “Playground” tab to access it.

Call Types

You can simulate two types of calls:

Type	Description	Use Case
📞 Inbound	Customer calling your agent	Test welcome flow, support scenarios
📤 Outbound	Agent calling a customer	Test introductions, sales scripts

Switch between call types to test both greetings and ensure your agent handles each scenario correctly.

Sending Messages

Text Input

Type messages in the input field and press Enter or click Send. The agent will respond as it would in a real call.

Voice Input

Click the microphone icon to speak your message. The agent will:

Transcribe your speech
Process the message
Respond with synthesized voice

Voice testing helps verify speech recognition accuracy and voice synthesis quality.

Conversation View

The Playground displays messages chronologically:

Element	Description
👤 User Message	Your inputs (text or transcribed voice)
🤖 Agent Message	Agent responses
🔧 Tool Call	When agent triggers a webhook or action
✅ Tool Result	Response from webhook or action

Audio Playback

For each agent response, you can:

▶️ Play the audio response
⏸️ Pause playback
🔁 Replay messages

Context Variables

Test how your agent handles different input parameters:

Open Context Variables

Click the “Context Variables” panel in the Playground.

Set Values

Enter test values for each input parameter.

Start Conversation

Begin testing with the configured context.

Example Context

{
  "customer_name": "John Smith",
  "order_number": "ORD-12345",
  "is_premium_customer": true,
  "account_balance": 1250.50
}

Reliability Test

Run multiple iterations of the same conversation to test consistency:

Configure Scenario

Set up your test conversation and context variables.

Set Iterations

Choose how many times to run the test (e.g., 10, 50, 100).

Run Test

Click “Run Reliability Test” to start.

Review Results

Analyze consistency across all iterations.

Reliability tests help identify edge cases where the agent might behave inconsistently.

Testing Features

Test Tool Calls

When your agent triggers a webhook:

The tool call appears in the conversation
You can see the parameters being sent
The response (or mock response) is displayed

Test Evaluations

After a conversation, you can run evaluations to verify they work correctly:

End the conversation naturally
Click “Run Evaluations”
Review which evaluations pass or fail

Test Knowledge Base

Verify your agent can find and use knowledge base content:

Ask questions that require KB lookup
Observe if the agent retrieves correct information
Check if responses are accurate and relevant

Best Practices

Test Edge Cases

Try unusual inputs, interruptions, and unexpected responses to see how your agent handles them.

Vary Context

Test with different input parameter combinations to ensure personalization works correctly.

Test Both Call Types

Always test both inbound and outbound scenarios.

Check Tool Integrations

Verify webhooks are called with correct parameters.

Run Reliability Tests

Periodically run multi-iteration tests to ensure consistency.

Common Testing Scenarios

Happy Path

Test the ideal conversation flow where everything goes as expected.

Error Handling

Test what happens when:

User provides invalid information
Webhooks fail
User doesn’t respond

Objection Handling

Test how the agent responds to:

“I’m not interested”
“Call me back later”
“I need to think about it”

Edge Cases

Test unusual scenarios:

Very short responses
Long, rambling inputs
Off-topic questions
Multiple questions at once

Troubleshooting

Issue	Solution
Agent not responding	Check if LLM model is configured correctly
Wrong greeting	Verify call type (inbound/outbound) is set correctly
Missing personalization	Ensure context variables are set
Webhook not triggering	Check tool descriptions and conditions
Slow responses	Try a faster LLM model

Getting Started

Agents

Advanced Editor

Voice Library

Calls

CRM

Account

What is the Playground?

Accessing the Playground

Call Types

Sending Messages

Text Input

Voice Input

Conversation View

Audio Playback

Context Variables

Example Context

Reliability Test

Testing Features

Test Tool Calls

Test Evaluations

Test Knowledge Base

Best Practices

Common Testing Scenarios

Happy Path

Error Handling

Objection Handling

Edge Cases

Troubleshooting

Next Steps

Copilot

Version History

Getting Started

Agents

Advanced Editor

Voice Library

Calls

CRM

Account

​What is the Playground?

​Accessing the Playground

​Call Types

​Sending Messages

​Text Input

​Voice Input

​Conversation View

​Audio Playback

​Context Variables

​Example Context

​Reliability Test

​Testing Features

​Test Tool Calls

​Test Evaluations

​Test Knowledge Base

​Best Practices

​Common Testing Scenarios

​Happy Path

​Error Handling

​Objection Handling

​Edge Cases

​Troubleshooting

​Next Steps

Copilot

Version History

What is the Playground?

Accessing the Playground

Call Types

Sending Messages

Text Input

Voice Input

Conversation View

Audio Playback

Context Variables

Example Context

Reliability Test

Testing Features

Test Tool Calls

Test Evaluations

Test Knowledge Base

Best Practices

Common Testing Scenarios

Happy Path

Error Handling

Objection Handling

Edge Cases

Troubleshooting

Next Steps