Skip to content

Sandbox

Welcome to the Accelerat.ai Sandbox environment! This is your pre-production testing space where you can validate your bot assistants, prompts, and the orchestrator before deploying them to production.

What is the Sandbox?

The Sandbox is a safe testing environment where you can:

  • 🧪 Test agents and bots before moving to production
  • 💬 Interact via chat with your bots and test their responses
  • 🎯 Validate prompts and fine-tune bot behaviors
  • 🔄 Test the orchestrator and verify the complete agent execution
  • 📊 Experiment safely without affecting live users

How to Use the Sandbox

1. Environment Access

To access the Sandbox environment:

  1. Create an agent
  2. Deploy the agent to a sandbox environment
  3. Click on the Chat tab

2. Testing Your Agents

When you create an agent with bots in the Sandbox, you can:

  • Test bot interactions: Use the integrated chat to interact with your bots
  • Validate prompts: Test and refine bot prompts to ensure optimal responses
  • Test orchestration: Verify how the orchestrator coordinates between different bots
  • Simulate scenarios: Create different conversation scenarios to validate behavior
  • Debug in real-time: Monitor bot responses and adjust configurations on the fly

3. Chat Interface

The Sandbox provides a chat interface where you can:

  • Start conversations with your configured bots
  • Test different user inputs and edge cases
  • Validate bot responses and prompt effectiveness
  • Verify the orchestrator's decision-making process
  • Test multi-bot interactions and handoffs

4. Sandbox Limitations

!!! warning "Important Limitations" - Data in the Sandbox is not persistent and may be reset periodically - The Sandbox environment has limited resources compared to production - Do not use real or sensitive data in the testing environment - Chat history may be cleared during resets

Usage Examples

Testing a Bot Agent

1. Create an agent with bots in the Canvas
2. Configure bot prompts and behaviors
3. Deploy the agent
4. Open the chat interface
5. Interact with the bot using different scenarios
6. Validate responses and adjust prompts as needed
7. Test the orchestrator's agent control
8. When satisfied, deploy to production

Validating the Orchestrator

1. Create the agent
2. Deploy the agent
3. Go to the Chat tab
4. Click on the Orchestrator agent on the sidebar
5. Chat with or call the orchestrator
6. Deploy to production with confidence

Best Practices

!!! tip "Tips for using the Sandbox" - ✅ Always test agents thoroughly in Sandbox before production deployment - ✅ Test multiple conversation scenarios and edge cases - ✅ Validate prompt effectiveness through chat interactions - ✅ Use realistic test data that mimics production scenarios - ✅ Test orchestrator logic with various user inputs - ❌ Don't skip Sandbox testing for critical agents - ❌ Don't use production credentials or real user data

Additional Resources

Support

If you encounter problems or have questions about the Sandbox environment, contact our support team through the help portal.

Ready to get started? Access the Sandbox and start experimenting! 🚀