Test and improve AI agent responses more efficiently

Until now, generated test questions ran automatically and consumed AI credits before you could review them. You also had to open each test case individually to see the AI agent's response.

With the improved AgentFlow testing experience, you can:

Review questions before running tests
: Check generated, imported, and manually added questions before you click Run, giving you more control over AI credit usage.
Review test results faster
: See a preview of each AI response and its "Good" or "Poor" rating directly in the batch testing table, without opening every case.
Understand how AI responds
: Use the "How it answered" option in chat testing to see which knowledge sources and steps your AI agent used to generate a response.
Review side-by-side
: Review individual test cases in a side panel without losing sight of the batch testing table.