Testing tools in the preview

The preview keeps the test widget's tools: restart the conversation, set session parameters, locate the dialogue node, see predicted intents and open the log.

The preview keeps the test widget's tools: restart the conversation, set session parameters, locate the dialogue node, see predicted intents and open the log.

Restart the conversation

Click Restart conversation in the toolbar. It does the same as the restart in the test widget: a new test conversation starts from the beginning. Use it also when the preview is empty or shows only part of a conversation.

Set session parameters

Click Session parameters in the toolbar to open the dashboard's own session parameters dialog. Session parameters exist only in a test of a dialogue, a branch or a procedure, which you start with Test dialogue, Test branch or Test procedure in the dialogue builder. In a test started with Test AI agent, the button is greyed out; see A test tool is greyed out or does nothing.

The values of the current test are listed in the inspector, on the Conversation tab, where you can also copy them. How the dashboard's test options work is in Testing an AI agent before you publish it.

Use the tools next to a message

Hover over a message to show its tools:

• Locate in dialogue builder (the crosshair) jumps to the dialogue node that sent the message. Like session parameters, it works in tests of a dialogue, a branch or a procedure.
• Show predicted intents (the three dots) appears on customer messages for which the AI agent predicted intents.
• Inspect message (the angle brackets) opens the message in the inspector. See Using the inspector.

To hide the tools, turn off Test tools on hover in the preview settings.

Open the conversation logs

Open in conversation logs opens the current test conversation in the conversation logs of the AI agents dashboard, in a new tab. Copy link to conversation log, just below it, copies the link to that conversation instead, so you can share it with a colleague who has access to the AI agent. Test conversations are recorded there, as Reviewing conversation transcripts explains. The same information is on the Log tab of the inspector; see Reading the conversation log.

Run a test set

Test sets in the toolbar keeps sets of customer messages per AI agent, with optional checks on the answers. Run starts a new test conversation, sends the messages one by one and compares the answers with the previous run. While a set runs, the preview is read-only and the toolbar button shows which message it is on. How you build and read a test set is in Testing an AI agent with test sets.

Leave the preview or make room

• Back to test widget turns the preview off and shows the test widget again.
• Close test closes the whole test, the same as the X at the top of the test widget. Preview mode stays on, so the next test opens as a preview again.
• Hide tools folds the toolbar in, leaving the pluscloud badge and Show tools. Click Show tools to open the toolbar again.

See also

• Using the inspector
• Testing an AI agent with test sets
• A test tool is greyed out or does nothing
• Testing an AI agent before you publish it
• Reviewing conversation transcripts