This is where the system finally starts working with what you’ve set up so far.

Select the dataset you want to test

Choose the Agent / Model config you’ve saved earlier and simply click Run Evaluation. Disseqt will instantly start simulating multiple conversational paths inside that dataset and begin processing the results.

Once the evaluation completes, you’ll be automatically redirected to the results overview screen where you’ll see pass/fail summaries, anomaly flags, hallucination signals, and stability indicators.

This is the moment where you start observing real performance patterns this data becomes your baseline before you move into refining safety or advanced checks.