For AI Agents explained: a practical look at testing, I would separate the problem into setup, output quality and repeatability. agents and practical fits into that because it can change the result quite a lot depending on the workflow. It also gives you a cleaner baseline if you want to compare...