Change one thing, then check it
AI Studio lets you change a rule and watch the macro’s next decision change with it, but only if you change one thing at a time. Change five things at once and you won’t know which one did it. This page is the loop that keeps a profile readable: change one thing, check it on a scenario, check it on a real gather, and remove what didn’t help.
Start from the decision you saw
Section titled “Start from the decision you saw”When the rules in your profile lead to decisions you think are incorrect, you can iteratively tune and test changes to it to reach your goals. Trace which rules or curves are incorrect, modify them, and test them live.
That starts with a decision you actually watched. If you don’t have one yet, Watch a gather and find one thing to change gets you one. Once you can say “when X, I’d rather the macro Y”, Which editor fixes what tells you which editor to open, and you make exactly one change there.
Check it in Simulator first
Section titled “Check it in Simulator first”Open Tools → Simulator and load the scenario closest to your moment. Early field, low buffs is a sparse field with Melody and Focus near empty, which makes a buff rule easy to read. Dense cluster bunches six tokens together with close scores, which makes a weight or distance change easy to read. Click Trace on the winner and find the line your change should have moved.
Try to predict the new score before you look at it. If it moved for the reason you expected, keep the change. If the winner didn’t change, or changed for a reason you didn’t expect, undo it and read Trace again before you try a second edit. The Simulator ranks ordinary tokens and actions. Duped tokens are scored separately, so you check Duped order on a real gather.
Then check it on a real gather
Section titled “Then check it on a real gather”While testing your profile live, your field and buffs can lead to decisions that can be incorrect, and a scenario can’t reproduce every one of them. Click Save in the file bar. When Revolution Macro is linked, it reloads the profile as soon as the save succeeds. Then click Gather in the session controls, watch for a few minutes, and click Stop. AI Studio opens the recording in Test Bench, where you can step to the moment that matters and read the same score breakdown for the real field.
If you’re going to make several small edits during one session, turn on Live Apply (the lightning control next to Save). It saves each valid edit as you make it, and the linked macro reloads it while gathering. Rules with validation errors are skipped until you fix them. Save and sync a movement profile covers the badges you’ll see.
Know when to stop
Section titled “Know when to stop”An over-tuned profile is one where you can no longer predict the winner before expanding Trace. The symptoms show up in the field first: the macro hikes across the field for tokens that aren’t worth the walk, because stacked add rules have grown scores large enough that the distance curve’s multiplier stops mattering. Tokens sit on the ground permanently Gated because a blocking gate written weeks ago still matches them. Buff timing feels random because three overlapping rules each nudge the same buff in different directions.
The tell inside AI Studio is a Trace you can’t read: a dozen rule contributions on every candidate, none of which you can attribute to a decision you remember making. When that happens, stop tuning forward and start removing. Disable recent rules one at a time in the Simulator until the ranking makes sense again, then re-add only the ones whose effect you can name. Every rule you keep should answer “which pick does this change, and why do I want that?” A profile where every rule has an answer is tuned. A profile where you’re not sure is over-tuned.