Why AI Gather
Traditional patterns like Bowl, Kettle, and Skillet have been the meta for years. They cover the field, pop bubbles with camera rotations, and fight drift with corner alignment, but they do not represent the upper ceiling of efficiency for blue hive honeymaking. AI Gather is different - it detects where tokens are physically on a field to then score and collect them. This page covers why that matters, what the macro can see, and what it can’t know.
Fixed routes commit too early
Section titled “Fixed routes commit too early”Not every token can be efficiently collected by a pattern, nor are patterns reactive to changes in abilities or buffs like the Pop Star or Tide Surge. Their movements are highly optimistic about the spawning and position of tokens, especially duped tokens which require the player to stand at specific points for a fixed amount of time. For red and white hives, patterns have been especially useless, as their honeymaking is centered around the collection of targets, which require precise, coordinated movements.
AI Gather detects where tokens are physically on the field and combines that with your buffs and abilities. The system is responsible for the detection of tokens and aggregation of buffs, abilities, and other data, while allowing you to customize which token or action is chosen at any given moment - just like you would playing manually. Think of it this way: AI Gather isn’t just another ‘pattern’ that runs fixed movements, but rather a framework that collects data and dynamically chooses what actions to take based on that data.
What the macro can see
Section titled “What the macro can see”Revolution Macro provides several AI models that are designed to detect tokens on the field, with different model types being optimized for their respective hive colors. While gathering, the macro predicts where tokens are from images captured of the Roblox window, many times a second. It keeps a memory of the tokens it has seen, so it knows how many times it has seen each one, how long it has been on the field, and when it was last visible. A token hidden behind a bubble for one frame doesn’t disappear from that memory, which is what lets the macro trust a token, guess when it will expire, and notice clusters.
That memory is what the score ranks, not one-off detections. Detection and movement run independently, so the macro can re-evaluate and retarget while your character is still walking, the way you would playing manually. The field view comes from the Roblox window, not from the game itself. If the macro misses a token that’s clearly on your screen, that’s a detection problem, and Detection problems is where to start, not a profile rule.
Why a score, not a script
Section titled “Why a score, not a script”Before AI Studio, AI Gather acted as a black box - very little could be customized besides base token weights. The scoring was mostly hard-coded and refined by the developers, but that restricted the ceiling for honeymaking. AI Studio gives you, the player, all of the control, by transforming AI Gather into a framework that delegates all of the scoring to profiles composed of priorities, rules, zones, curves, and more.
While gathering, those pieces are chained together to determine which token will be collected or what action will be executed, and each effect is directly observable in the Simulator and the Test Bench by clicking Trace. While you develop a profile, you’ll often be wondering questions like:
- Why did this token get scored the highest?
- Which rule, zone, or curve bumped up that token’s score the most?
- Why did a certain token get blocked?
AI Studio gives you the tools to break those questions down and see directly how your rules affect the macro’s decisions. Token detection is handled by AI vision models, but the decision itself can be traced and debugged. It isn’t an opaque model that just “does AI Gather”.
What a profile can’t do
Section titled “What a profile can’t do”A movement profile decides what to do with the tokens the macro can see. It doesn’t change what the macro sees or when it gathers. It can’t fix a detection problem, choose which fields to farm, or guarantee that one token is always collected, because scoring is ranking under real constraints like distance and expiry, and even a heavily favored token loses sometimes. That’s the system working, not failing.
To turn AI Gather on for a field, see Set up AI Gather. To watch the live decision loop, continue to How AI Gather decides where to move. To change the rules behind those decisions, start with About AI Studio.