What’s the best way to set up a bot test canvas for accurate results? or Has anyone tried using a bot

If your bot test canvas feels rigid, try adding some randomness. Like, throw in 10% wildcard inputs to see how it handles chaos.

I use Rasa’s testing framework for this—it’s got great tools for stress-testing.

Also, edge cases? I document ’em as they pop up and add ’em to the next test cycle.