Playground Mode Overview
Playground Mode allows you to simulate AI agent conversations without affecting live campaigns. It’s a safe environment to test scripts, prompts, voice settings, and knowledge base integrations.
1. Purpose of Playground Mode
- Validate conversation flow and agent behavior.
- Test prompt responses and branching logic.
- Check knowledge base retrieval and dynamic variable replacement.
- Evaluate voice, accent, and tone in a controlled setting.
- Identify errors, latency, or unnatural responses before live deployment.
2. Using Playground Mode
- Navigate to Agents → Playground.
- Select the agent you want to test.
- Choose a scenario or start a new conversation.
- Enter simulated lead responses or allow the agent to generate them automatically.
- Monitor the transcript, agent responses, and knowledge base lookups in real-time.
- Make adjustments to prompts, scripts, or settings as needed.
3. Features of Playground Mode
- Step-through simulation: Move through scripts step by step to verify branching logic.
- Variable injection: Test dynamic insertion of lead data (e.g.,
{lead_name}, {product}).
- Voice preview: Hear the agent speak in the configured language, accent, and tone.
- Error logging: Identify errors, unhandled responses, or failed knowledge base lookups.
4. Best Practices
- Start with simple scripts before testing complex branching.
- Test edge cases to ensure robust agent behavior.
- Use multiple test scenarios to cover varied lead responses.
- Refine prompts and scripts based on observations before moving to live calls.
Playground Mode is the safest way to iterate quickly. Always test changes here before deploying to live campaigns to avoid costly mistakes and maintain lead engagement.
