Test Logic (Web chat)

Iteratively test your AI agent while building using the "Test Logic" web chat experience in the Regal builder.

Web chat is useful for testing your agent while building, because it lets you test your agent before hooking up a phone number and routing, and specifically debug function calling (i.e.agent actions). The feature lets you easily rewind to earlier parts of a conversation and take it in a new direction to iteratively test different potential conversational paths. However, it does not support testing all functionality, and final E2E testing should be done through actual SMS or phone calls.

Test Dynamic Variables

You can add test values for your dynamic variables to see how the agent would use them in the course of the conversation. Any dynamic variables you refer to in the Agent Prompt are automatically added to this modal for you to add test values.

Action Invocations & Results

In the Test Logic UI you can see when an action is invoked, what payload was sent, and what result was returned to the AI agent so you can see what the agent can make use of in the conversation. This is useful for constructing and debugging custom actions.

See Action Result details:

✅ What you can test with "Test Logic" feature

• Conversational logic e.g., how it follows your task under different scenarios or objection/question handling or RAG responses
• Agent actions
• Personalization using test values for dynamic variables

❌ What you cannot test with "Test Logic" feature

• Unsubscribe
• Delivery
• E2E Latency over SMS or phone calls
• Transfers
• Routing
• Dispositions

▶️ VIDEO TUTORIAL