Skip to main content
The Playground is a testing environment where you can interact with your chat agent, edit prompts, adjust AI settings, and see how changes affect responses in real-time. It’s the perfect tool for refining your chat agent before deploying it to your website. Agent Playground

​
Prerequisites

Before you begin, ensure you have:
  • An account and are logged in
  • An organization set up
  • A project created
  • Permission to update the project (you have this if you’re the project creator or admin)

​
What is the Playground?

The Playground helps you:
  • Test your chat agent by having real conversations with it
  • Edit prompts and see how changes affect responses
  • Adjust AI settings like model, temperature, and tools
  • Experiment safely without affecting your live chat agent
  • Save changes to your project when you’re satisfied
The Playground uses your project’s current settings but allows you to test custom configurations before saving them.

​
Access the Playground

1

Navigate to your project

After logging in, go to the Projects page from the main navigation. Find and click on the project you want to test. You’ll be taken to the project dashboard.
2

Open Playground

Look for Playground in the project sidebar or navigation menu. Click on it to expand the menu, then click Chat Playground. The Playground page will open with a chat interface on the right and configuration panel on the left.

​
Understanding the Playground Interface

The Playground is divided into two main sections:

​
Left Panel: Configuration

This panel lets you:
  • Select and edit prompt templates
  • Adjust AI model settings
  • Configure tools and actions
  • Save your changes

​
Right Panel: Chat Interface

This panel shows:
  • Your conversation with the chat agent
  • Messages you send and responses you receive
  • Real-time testing of your configurations

​
Test Your Chat Agent

1

Start a conversation

Once the Playground loads, you’ll see a welcome message. Type a question or message in the input field at the bottom of the chat interface. For example, try asking: “What can you help me with?”
2

Send your message

Click the send button or press Enter to send your message. The chat agent will respond based on your current prompt and AI settings.
3

Review the response

Read the chat agent’s response carefully. Consider:
  • Does it answer your question correctly?
  • Is the tone appropriate?
  • Does it use information from your knowledge base (if configured)?
  • Is the response helpful and relevant?
Try asking different types of questions to test various scenarios. Test both simple and complex queries to see how your chat agent handles them.
4

Continue the conversation

Keep asking questions to test different aspects of your chat agent. The conversation history is preserved, so you can have extended dialogues to test context understanding.

​
Edit Your Prompt

The prompt (also called “Agent Template”) tells your chat agent how to behave. You can edit it in the Playground to improve responses.
1

Select a prompt template

In the left panel, find the Agent Template (Chat) section. Click the Select Template dropdown to see available prompt templates. Each template has a name and description to help you choose the right one.
Templates are pre-configured prompts designed for specific use cases like customer support, sales, or FAQ assistance.
2

View the prompt content

After selecting a template, you’ll see the prompt content in a text editor below. This is the instructions that tell your chat agent how to behave.
3

Edit the prompt

Click in the text editor and modify the prompt to customize it for your needs. You can:
  • Change the chat agent’s role or personality
  • Adjust the tone (professional, friendly, casual)
  • Add specific instructions about what to do or avoid
  • Include examples of good responses
  • Reference your knowledge base or products
When you edit a predefined template, the system automatically creates a custom version so you don’t lose the original template.
4

Preview your prompt (optional)

Click the Preview button to see how your prompt will look when formatted. This helps you check if your prompt is well-structured.
5

Test the changes

After editing the prompt, go to the chat interface and ask a question. The chat agent will use your edited prompt to respond. Compare the response to previous ones to see if your changes improved it.
Changes to the prompt are not saved automatically. You’ll need to click “Save Settings” to apply them to your project.

​
Adjust AI Settings

You can modify AI configuration settings in the Playground to see how they affect responses.

​
Change the AI Model

1

Find the model setting

In the left panel, look for the Model dropdown in the AI Configuration section.
2

Select a different model

Click the dropdown and choose a different AI model. Common options include:
  • Gemini 2.0 Flash: Fast and efficient, great for most real-time interactions
  • gpt-4o-mini: Cost-effective, good for standard use cases
  • gpt-4: More powerful, better for complex tasks
Start with gpt-4o-mini for most chat agents. Only use more powerful models if you need advanced capabilities.
3

Test the new model

Ask a question in the chat to see how the new model responds. Different models may have different response styles or capabilities.

​
Adjust Temperature

Temperature controls how creative or focused your chat agent’s responses are.
1

Find the temperature setting

In the left panel, look for the Temperature slider or input field.
2

Set the temperature

Adjust the temperature value:
  • Lower values (0-0.3): More focused, consistent, predictable responses
  • Higher values (0.7-1.0): More creative, varied, conversational responses
Use lower temperature (0.3) for customer support where accuracy is important. Use higher temperature (0.7) for more conversational or creative chat agents.
3

Test different temperatures

Try asking the same question with different temperature settings to see how it affects the response style.

​
Configure AI Tools/Actions

AI Actions are special capabilities that extend your chat agent beyond basic conversation.
1

Find the AI Actions section

In the left panel, look for the Available Tools & Actions (Chat) section. You’ll see a list of available tools with checkboxes.
2

Enable tools

Check the boxes next to the tools you want your chat agent to use. Common tools include:
  • Knowledge Base Search: Allow the agent to search your uploaded documents to answer questions
  • Book Appointment: Schedule appointments (requires Calendly setup)
  • Track Shipment: Track shipments (requires Shiprocket setup)
  • Human Handover: Transfer conversations to human agents
Some tools require additional setup (like credentials). If you check a tool that needs setup, a modal will appear to guide you through the configuration.
3

Configure tool credentials (if needed)

If a tool requires credentials:
  • A modal will appear when you check the tool
  • Follow the instructions to enter your credentials
  • The system will validate them before enabling the tool
Only enable tools that you actually need. Each tool adds functionality but may require additional setup and configuration.
4

Test the tools

After enabling tools, test them in the chat. For example, if you enabled “Book Appointment,” try asking the chat agent to schedule a meeting.

​
Save Your Changes

After testing and adjusting your settings, you can save them to your project.
1

Review your changes

Before saving, make sure you’re happy with:
  • The prompt content and how it affects responses
  • The AI model selection
  • The temperature setting
  • The enabled tools
The Playground tracks unsaved changes. You’ll see indicators if you have changes that haven’t been saved yet.
2

Click Save Settings

In the left panel, find the Save Settings button (usually at the bottom of the configuration panel). Click it to save your changes to the project.
3

Wait for confirmation

You’ll see a loading indicator while your settings are being saved. Once complete, you’ll see a success message confirming that your changes have been saved.
Your changes are now applied to your project and will be used by your live chat agent.

​
Reset Your Session

If you want to start a fresh conversation in the Playground:
1

Find the reset button

Look for a reset or clear button (usually an icon with circular arrows) in the chat interface.
2

Reset the conversation

Click the reset button to clear the conversation history and start a new session. This doesn’t affect your saved settings - it only clears the chat messages.

​
Best Practices

​
Testing Your Chat Agent

  • Test various question types: Ask simple questions, complex questions, and edge cases
  • Try different phrasings: See how the chat agent handles the same question asked in different ways
  • Test knowledge base integration: If you have knowledge sources, ask questions that should be answered using them
  • Check tool functionality: If you’ve enabled tools, test them to ensure they work correctly
  • Review tone and style: Make sure responses match your brand voice

​
Editing Prompts

  • Start with templates: Use pre-made templates as a starting point - they’re often better than starting from scratch
  • Make small changes: Edit prompts incrementally and test after each change
  • Be specific: Clearly describe what your chat agent should do, what tone to use, and what to avoid
  • Include examples: Add examples of good responses in your prompt to guide the AI
  • Test thoroughly: Always test prompt changes before saving them

​
Adjusting AI Settings

  • Start with defaults: Begin with default settings and adjust based on testing results
  • Change one thing at a time: Modify one setting at a time so you can see what each change does
  • Document your changes: Note which settings work best for your use case
  • Consider costs: More powerful models may cost more, so balance performance with cost

​
Using Tools

  • Only enable what you need: Don’t enable tools you won’t use - they add complexity
  • Read tool descriptions: Understand what each tool does before enabling it
  • Configure properly: Make sure to complete credential setup for tools that require it
  • Test after enabling: Always test tools after enabling them to ensure they work

​
Troubleshooting

Possible reasons and solutions:
  • WebSocket connection failed: Refresh the page and try again
  • Project not found: Make sure you’re accessing the correct project
  • Network issues: Check your internet connection
  • Browser issues: Try using a different browser or clearing your browser cache
What to check:
  • Verify that the connection status shows “Connected”
  • Check if there are any error messages in the chat
  • Make sure you’ve saved your prompt and AI settings
  • Try refreshing the page and reconnecting
  • Check that your project has valid AI configuration
What to do:
  • Make sure you’ve clicked “Save Settings” after making changes
  • Refresh the page to ensure the latest settings are loaded
  • Check that you’re testing with the correct prompt template selected
  • Verify that your changes were actually saved (check for success message)
  • Try resetting the conversation and asking a new question
What to check:
  • Make sure you have permission to update the project
  • Verify that a prompt template is selected
  • Try selecting a different template and then editing it
  • Check if there are any error messages in the browser console
  • Refresh the page and try again
What to do:
  • Verify that tool credentials are configured correctly
  • Check that the tool is actually enabled (checkbox is checked)
  • Test the tool with a specific request (e.g., “Book an appointment”)
  • Review the tool’s description to understand what it does
  • Check if there are any error messages when using the tool
What to check:
  • Make sure you have permission to update the project
  • Verify that you’ve made actual changes (not just viewing settings)
  • Check your internet connection
  • Look for any error messages
  • Try refreshing the page and saving again

​
Tips for Success

​
Iterative Testing

  • Test, edit, test again - make small changes and test frequently
  • Keep a record of what works and what doesn’t
  • Compare responses before and after changes
  • Ask team members to test and provide feedback

​
Prompt Engineering

  • Be clear and specific in your instructions
  • Define the chat agent’s role and personality
  • Set boundaries (what it should and shouldn’t do)
  • Provide context about your business or products
  • Include examples of good responses

​
Performance Optimization

  • Start with simpler settings and increase complexity as needed
  • Monitor response quality vs. cost (especially with different models)
  • Test temperature settings to find the right balance
  • Only enable tools that add real value

​
Collaboration

  • Share your Playground session with team members for feedback
  • Document successful prompt configurations
  • Keep notes on what settings work best for different use cases
  • Use the Playground to demonstrate chat agent capabilities to stakeholders

​
Summary

Using the Playground involves:
  1. Access the Playground from your project
  2. Test your chat agent by having conversations
  3. Edit prompts to improve responses
  4. Adjust AI settings (model, temperature, tools)
  5. Test changes in real-time
  6. Save your changes when satisfied
The Playground is your testing and refinement environment. Use it to experiment, iterate, and perfect your chat agent before deploying it to your website. Take your time to test thoroughly - the better you refine your chat agent in the Playground, the better it will perform for your users.

​
Quick Reference

Playground Features:
  • Real-time chat testing
  • Prompt editing and customization
  • AI model and temperature adjustment
  • Tool/Action configuration
  • Save changes to project
Key Actions:
  • Test: Ask questions in the chat interface
  • Edit: Modify prompts in the text editor
  • Adjust: Change AI settings in the configuration panel
  • Save: Click “Save Settings” to apply changes
Time to Test: 10-30 minutes (depending on how thorough you want to be) Can You Edit Later? Yes! You can return to the Playground anytime to test and make further adjustments.