Excited to share experiments in @mastra. Create test cases, score the results, and compare runs side by side to see if the changes actually helped. Swap a model, tweak a prompt, run again, and compare. Do it all in Studio or via the API. 1/5↗