Playground
Test your AI chatbot before deploying it. The Playground is your sandbox for testing. Ask questions in natural language, see the SQL LILA generates, and verify that data isolation works correctly. All testing happens here without affecting your live widget or production data.
Navigation: Dashboard → Select Project → Playground
Connection Status
The Playground automatically connects to your project’s AI engine when you open the page. The connection indicator in the header shows:
- Connected (green): Ready to test queries
- Disconnected (gray with spinner): Connecting to AI engine
- Project Paused (red): Playground unavailable until project is activated
New Session
Click New Session to start a fresh conversation and clear all previous messages. This resets the chat history but preserves your test configuration settings.
Test Configuration
Test different user scenarios by overriding your project’s default settings for testing purposes. Changes made here apply only to the current Playground session and are not saved to your project.
User Simulation
Simulate different users and roles to test data access controls:
- User ID (required): Simulates a specific user from your application
- Tenant ID (required when Multi-Tenant enabled): Tests organization-level data isolation
- Role: Tests role-based access control (e.g., admin, user, manager)
Data Isolation
Configure how queries filter data:
Multi-Tenant
- Toggle on to filter queries by tenant/organization
- Specify the column name that stores tenant IDs (e.g.,
client_id,org_id) - When enabled, Tenant ID becomes required in User Simulation
User Isolation
- Toggle on to filter queries by individual users
- Specify the column name that stores user IDs (e.g.,
user_id,owner_id) - Restricts query results to data owned by the simulated user
Currency
Single Currency toggle determines how the AI formats monetary values:
- On: All prices displayed with one currency symbol (specify USD, EUR, AED, etc.)
- Off: AI handles multiple currencies from your database
Context
Help the AI understand your data better:
Tool
- Select the platform your data comes from (Shopify, WooCommerce, Custom, etc.)
- Improves AI’s understanding of standard fields and relationships
Industry
- Select your business vertical (E-commerce, Healthcare, SaaS, etc.)
- Helps AI interpret industry-specific terminology
Business Context
- Add custom rules, terminology, or special conditions
- Example: “VIP customers have loyalty_tier > 3. Sales tax is stored in tax_amount column.”
Chat Interface
Test queries in real-time with streaming responses:
- Type your question in the message box
- Press Enter to send (Shift+Enter for new line)
- Watch the AI response stream in real-time
- Review the generated SQL and performance metrics in Debug Info
Suggested Test Queries:
- “What were yesterday’s sales?”
- “Show me top customers this month”
- “How many orders are pending?”
Clear Chat button removes all messages from the current session.
Debug Info
View the SQL LILA generates in the debug panel. Technical details for each query include:
- SQL Generated: The actual database query the AI created
- Execution Time: How long the query took (in milliseconds)
- Tokens Used: Number of AI tokens consumed
- Model: AI model used for generation
- Retries: Number of self-healing attempts if the initial SQL failed
- Filters Applied: Data isolation filters automatically added (tenant, user, etc.)
Use this information to optimize your configuration and understand how the AI interprets questions.
Paused Projects
When a project is paused, the Playground displays an overlay message and blocks all testing. Activate the project from Settings → Setup & Integration to resume Playground access.
Related Pages
- Project Settings - Configure default values loaded in Playground
- Business Logic - Define rules and access controls tested here
- Schema Upload - Database structure the AI queries against