Run a Test

Execute your scenario against an agent and watch the results come in.

Start a run

A run executes all the tests in your scenario against a specific agent.

On the scenario detail page, click Run in the top-right corner.
New run opens, filled in with your scenario and all of its enabled tests.
Select an agent - this is the chatbot you want to test. If you don’t have one connected yet, use a Demo agent to try things out.
Leave the personality at Neutral, or pick another one. Neutral uses straightforward, standard communication; you can create custom personalities later - see Personalities.
Click Start run.

After you click Start run, you’re taken to the run page. Voxli starts executing each test in your scenario one by one.

Each test shows its current status: running, passed, failed, or canceled.
Results appear as each test completes. You don’t need to wait for all tests to finish.
The score at the top of the page tracks the average across finished tests, and the counts beside it track how many hit a blocker or a warning; click a count to filter the table down to just those.

Each row in the results table shows:

A status dot: running, passed, failed, or canceled.
Test - the name you gave the test.
Score - the test’s score as a percentage. Hover it to see how many blocker, medium, and low assertions passed.

If your run covers more than one scenario, personality, or agent, the table adds a column for that too.

Click any test row to open its test result page with the full conversation and assertion outcomes. See Results for a deep dive.

Results - learn how to inspect conversations and read assertion outcomes.
Advanced Run Options - select specific tests, use multiple personalities, or run tests multiple times to detect flaky behavior.
Personalities - create custom communication styles for the AI tester.