The Preview feature allows you to test how your AI Agent behaves before going live.

You can simulate conversations, validate your setup, and ensure your Scenarios and instructions work as expected.

‍

What you can test

Responses and tone

Use Preview to evaluate how your AI Agent communicates:

  • Check how the AI Agent responds to different questions
  • Validate tone of voice and clarity
  • Ensure answers align with your knowledge base and instructions

Scenarios and workflows

Preview helps you confirm that your automation behaves correctly:

  • Test whether the correct Scenario is triggered
  • Verify step-by-step execution of workflows
  • Check if the AI Agent asks for the right information at the right time

Actions inside conversations

Your AI Agent can simulate actions during a conversation, such as:

  • Adding internal comments
  • Assigning conversations to a team or user
  • Applying labels

In Preview, these actions are displayed visually so you can easily verify that everything works as intended.

Beispiel:

‍
You might see a generated internal comment or a message like “Handed over to [team]”. This confirms that your Scenario instructions are being executed correctly.

Tips for effective testing

Test complete flows

Avoid testing single messages in isolation.
Simulate full conversations to validate the entire workflow from start to finish.

Try different variations

Test how your AI Agent handles different inputs:

  • Rephrase the same question
  • Use incomplete or unclear messages
  • Explore edge cases

This ensures your AI Agent performs well in real-world situations.

Test Scenarios explicitly

Go through each Scenario you’ve created and verify:

  • Is it triggered at the right moment?
  • Does it follow all defined steps?
  • Are actions executed correctly?

Test with images

If your AI Agent supports image handling:

  • Upload relevant images (e.g. a damaged product)
  • Check whether the AI Agent interprets and uses the image correctly in its response

Validate handover behaviour

Make sure escalation works as expected:

  • Test when and how handovers are triggered
  • Confirm the correct team or user is assigned
  • Verify that all required information is collected before handover

Summary

Preview helps you to:

  • Test responses, tone, and knowledge
  • Validate Scenarios and workflows
  • Confirm actions like assignments and internal comments

Use Preview to refine your AI Agent and ensure it performs as expected before going live.