Two Modes
Embedded Playground
Open it from within the AIP editor by clicking the Preview button in the toolbar. A testing panel opens alongside your editor. Best for: Iterating on a specific AIP — edit instructions, test, adjust, repeat.- Brand and AIP are automatically set to match what you’re editing
- Uses draft versions so you see your latest changes immediately
- Your conversation persists while you switch between editing and testing
- Close and reopen without losing your chat history
Standalone Playground
Navigate to Workspaces → Playground for the full-page testing environment. Best for: End-to-end testing before going live — test the complete player experience across AIPs, brands, and configurations.- Full control over all settings (brand, agent, AIP, channel, locale)
- Test classification by sending messages without pre-selecting an AIP
- Compare behavior across different brands and agents
Configuration
Settings can only be changed before starting a conversation or after starting a new chat. Once a conversation is active, settings are locked to ensure consistent behavior.
Brand
Select which brand to test against. This affects:- Which brand variant of AIPs is used
- Brand-specific Knowledge Base articles
- Brand-specific backoffice data
Agent
Choose which AI agent persona to use. Useful for testing different tones, styles, and agent-specific instructions.AIP Selection
Skip classification and jump directly to a specific AIP — even inactive or draft ones. This lets you:- Test draft AIPs before publishing
- Test inactive AIPs before activating
- Test specific procedures without needing to phrase the right trigger message
Channel
Switch between Chat and Email modes:- Chat: Conversational, shorter responses
- Email: More formal, complete responses with proper email formatting
Locale
Test how the AI agent handles different languages and regions by selecting a locale.Preview Mode (Draft vs Published)
Toggle between testing draft (unpublished) and published versions of your AIPs. The Playground defaults to draft mode so you can test changes before they go live.Player Profiles
Player profiles let you save and reuse sets of authentication fields — so you don’t have to re-enter player details every time you start a new test.Creating a Profile
Fill in auth fields
Enter the authentication fields your integration requires (email, player ID, session token, etc.)
Using Profiles
Select a saved profile before starting a conversation. The profile’s auth fields are sent with the initial message, simulating an authenticated player.Profiles are saved locally in your browser. They persist across sessions but are not shared between devices or team members.
Testing with Backoffice Data
Mock Backoffice
Enable Use Mock Backoffice to test without hitting your real back-office APIs. Describe the scenario you want to simulate in natural language:Real Backoffice
Disable mock mode to use your actual back-office integration. Combine with a player profile to pre-authenticate and fetch real player data during the test.Select or create a player profile
Enter the auth fields your integration requires (email, session ID, etc.)
During a Conversation
Once a conversation is active:- Settings are locked — You cannot change the brand, agent, AIP, channel, locale, or backoffice mode. This ensures the conversation context stays consistent.
- Profiles can be managed — You can create and edit profiles, but changes won’t affect the current conversation.
- File attachments work — Send images and files just like a player would.
- Real-time responses — See typing indicators and receive responses as they happen.
Tips
Start with the embedded playground — When building or editing an AIP, use the sidebar preview. It’s the fastest way to iterate. Use the standalone playground for go-live testing — Before deploying to real players, run through your complete flows in the full-page playground. Save profiles for common scenarios — Create profiles for different player types (VIP, new player, restricted account) so you can quickly switch between them. Test classification — In the standalone playground, leave the AIP selector empty and send messages to test whether the right AIP gets triggered. Test edge cases — Try ambiguous requests, frustrated tones, incomplete information, and multiple topics in one message. Test both channels — Switch between Chat and Email modes to verify responses are appropriate for each format.Common Issues
| Issue | Cause | Fix |
|---|---|---|
| AIP not triggering | Description doesn’t match test message | Select the AIP directly, or refine the AIP description/example utterances |
| Old behavior showing | Testing with published version | Ensure preview mode is enabled (draft) |
| Player Context empty | Not authenticated or backoffice error | Use a player profile with valid auth fields, or enable mock backoffice |
| Settings won’t change | Conversation is active | Start a new chat — settings lock during active conversations |
| Wrong AIP selected | Multiple AIPs match the input | Make AIP descriptions more distinct, or select the AIP directly |
Related Documentation
- Testing AIPs — Overview of testing approaches
- AI Procedures (AIPs) — Creating and configuring AIPs
- Brand Variants — Brand-specific behavior
- Player Authentication — Setting up auth fields