Skip to content
Contact us
  • There are no suggestions because the search field is empty.

Playground

The Playground is a testing environment where you can safely try out your AI agents before letting them interact with real customers. It allows you to verify that your agent responds correctly, uses tools properly, and handles different scenarios as expected.

⚠️ Admin Access Only – Only users with admin permissions can access the Playground.

Purpose of the Playground

The Playground lets you:

  • Test if the AI gives the right answers.

  • Verify that the AI uses integrated tools correctly.

  • Ensure the AI handles greetings, transfers, or other actions appropriately.

  • Identify and fix mistakes before the agent interacts with real customers.

Think of it as a sandbox for safe experimentation.

Testing Modes

The Playground supports three testing modes: Chat, Live Conversation, and Outbound Call.

1. Chat

Use Chat mode to simulate a customer interacting with the AI via text.

  1. Go to the top navigation bar and click Playground.

  2. Select the Chat option.

  3. A chat window will open. Type messages as if you were a customer.

    • Example: "Hello, I need help with my order."

  4. Observe the AI’s responses.

    • Example: "I can certainly update your order with ID 045954059"

  5. Continue the conversation to test various scenarios:

    • Asking about products

    • Requesting human assistance ("I want to speak to a human")

    • Handling multiple questions in one session

2. Live Conversation (Voice)

Use Live Conversation to simulate spoken interactions with your AI agent.

  1. Go to the top navigation bar and click Playground.

  2. Select the Live Conversation option.

  3. A transcript window will open. Speak as if you were a customer.

    • Example: "Hello, I need help with my order."

  4. Listen to the AI’s spoken response.

    • Example: "Welcome! I’m Jane"

  5. Test different scenarios to ensure the AI responds appropriately, handles interruptions, or escalates to a human when needed.

3. Outbound Call

Use Outbound Call mode to test the AI’s ability to initiate calls to customers.

  1. Go to the Playground section.

  2. Select the Outbound Call option.

  3. Enter the phone number you want the AI to call (e.g., your own number for testing).

  4. Select whether the Agent should start the conversation or wait for the caller to answer first.

  5. You can also add custom parameters to pass extra information to the AI agent. These parameters use a key-value format and allow the AI to tailor the conversation based on context or customer-specific data.

  • Enter the Key – the name of the parameter.

    • Example: order_number, customer_name, preferred_language

  • Enter the Value – the data associated with that key.

    • Example: "12345" for order_number

    • Example: "John" for customer_name

  • These parameters can then be referenced in the AI’s prompt during the call.

  1. Click Start Call to initiate the test.

  2. Answer the phone and interact with the AI as a customer.

    • Example: "Hello, I’m calling to confirm your order."

  3. Use this mode to ensure the AI handles calls smoothly and responds naturally.

Best Practices

  • Test multiple scenarios: Try different questions, complaints, or requests to see how the AI handles them.

  • Iterate and adjust: If the AI makes mistakes, adjust its settings (e.g., prompt, variables, or data collection fields) and test again.

  • Simulate real conversations: Use realistic language, slang, or interruptions to ensure the AI is robust in real customer interactions.

  • Document issues: Keep notes of errors or unexpected behaviours to refine the agent’s configuration.

Using the Playground helps you ensure your AI agent is fully prepared before going live with it. If it makes mistakes, you can change its settings (like the Prompt) and test again.