The Make Anything Machine (MAM) represents the pinnacle of automation engineering in shapez 2. Unlike a traditional factory that produces one specific shape repeatedly, a properly designed MAM reads the Hub delivery goal in real time, decomposes the target signal into its constituent parts, routes raw shapes through configurable processing lanes, and assembles the final output — all without any manual reconfiguration. Building a MAM is not required to complete shapez 2, but it transforms the late game from an active factory management exercise into a self-sustaining automation system. This design guide covers the MAM architecture, core component roles, the signal-layer logic that makes dynamic production possible, wire routing strategies, and the key architectural decisions that separate a fragile prototype from a robust universal factory.

What Is a MAM and Why Design Matters

At its core, a MAM is a signal-driven factory. The Hub broadcasts a target shape signal that describes exactly what to build — which quadrants go where, what colors to apply, and how many layers to stack. The MAM reads this signal using a Global Receiver, then uses a cascade of simulated buildings (Unstacker, Rotator, Shape Analyzer) to decompose it into individual pin signals. These pin signals control Belt Filters that route the correct physical shapes into the correct processing lanes. The result is a factory that automatically reconfigures itself whenever the Hub goal changes.

The difference between a working MAM and a reliable MAM comes down to design discipline. A fragile MAM works for the shapes you tested it with but fails silently when it encounters a new target. A well-designed MAM handles every shape the Hub can request — it normalizes quadrants, checks for null signals, provides explicit trash paths for rejected shapes, and validates output against the target before delivery. The following sections break down the components, signal layers, and architectural patterns that enable robust MAM design.

MAM Component Types — Roles and Design Tips

A MAM is composed of several distinct component categories, each serving a specific role in the signal-to-shape pipeline. Understanding what each component does and where it belongs in the architecture is the first step toward designing your own MAM rather than copying someone else's.

The Five-Layer Signal Logic System

The most important concept in MAM design is the separation between the signal layer and the physical layer. The signal layer uses simulated buildings to operate on shape signals — deciding what should be built — while the physical layer uses real buildings to execute those decisions. A well-designed MAM keeps these two layers cleanly separated: signals flow from top to bottom through five distinct stages, and each stage is independently testable before the next stage is connected.

The five-layer architecture provides a framework for thinking about MAM design systematically. Layer 1 receives the target signal. Layer 2 decomposes it into manageable pin signals. Layer 3 matches those signals to physical shapes and routes them. Layer 4 executes the physical transformations. Layer 5 validates the output against the original target. Every MAM, regardless of its physical layout, follows this same signal flow pattern. The differences between architectures are in how Layers 3 and 4 are physically arranged — not in what the layers do.

A critical design rule for signal layers: place a Display at every handoff point between layers. The signal chain should be: Global Receiver → Display → Layer 2 modules → Display → Belt Filter inputs → Display → Physical modules. If you cannot see the signal at each stage, you cannot debug it when something goes wrong.

Wire System for Shape Routing

The wire system is the nervous system of a MAM. Pin signals travel along wires from the Signal Analyzer to every Belt Filter, Rotator controller, and Comparator in the factory. Unlike belts that carry physical shapes, wires carry information instantaneously — a signal change at the Global Receiver propagates to every connected component in the same tick. This instant propagation is what enables a MAM to switch production targets on the fly.

Wire routing requires careful planning because wires and belts occupy the same surface layer. The standard solution is wire tunnels — underground conduits that carry signals beneath your production lines, emerging only at connection points. Route all pin signals through tunnels so that the surface remains clear for belts carrying physical shapes. A common beginner mistake is routing wires on the surface in a way that blocks future belt expansion — by the time you need that space, the wire network is too complex to move.

Color-code your wire network by quadrant. Use one wire color for NE quadrant signals, another for NW, and so on. When debugging, you can instantly see which quadrant a signal belongs to without tracing the wire through multiple tunnels. This simple convention saves hours of debugging time in complex multi-lane MAMs.

Linear vs Branching Architectures

The physical layout of a MAM — how processing lanes are arranged — falls into three broad architectural patterns. Each has distinct trade-offs in throughput, complexity, and scalability. Choosing the right architecture for your current skill level and production goals is more important than copying the most advanced design you can find.

MAM Optimization Tips

• Start small — one layer, one NE quadrant. Your first MAM should process a single quadrant of a single-layer shape. Get the signal chain working end-to-end before adding complexity. A MAM that works for one quadrant can be expanded to four; a half-built MAM that tries to handle everything at once will never work at all.
• Normalize all quadrants to NE before comparison. Use Simulated Rotators to rotate every quadrant signal to the NE orientation. This ensures that the same shape in a different quadrant position produces the same comparison result — a circle in the SW quadrant matches a circle in the NE quadrant after rotation. Without normalization, your comparison logic produces false negatives and routes shapes incorrectly.
• Add trash paths at every rejection point. Every Belt Filter has a rejection output. Every Comparator has a mismatch path. Every processing lane has overflow. Connect every single one of these to a Trash building. A MAM that cannot dispose of wrong shapes will eventually deadlock when the Hub goal changes and the pipeline contains shapes from the previous target.
• Test with at least 5 different target shapes. A MAM that works for square shapes may fail for circles, pinned shapes, or multi-layer targets. Test your MAM against a diverse set of Hub goals — include at least one pinned target and one multi-layer target — before declaring it complete. The shapes that break your MAM will teach you more than the ones that work.
• Blueprint processing lanes, not the signal network. The physical processing modules (Cutter, Rotator, Painter, Stacker) are highly tileable and perfect for blueprinting. Save each processing lane as a reusable module. The signal routing — Global Receiver, Signal Analyzer, wire tunnels — is unique to each MAM layout and should be built manually per instance.
• Handle null signals explicitly. Null signals cause shapes to stall instead of routing to trash or output. Add null-check logic at every decision point: if a signal is null, send the shape to a dedicated reject path rather than letting it occupy a belt slot indefinitely. Use a Display to verify that null signals are being caught.
• Keep paint production centralized. Instead of building dedicated Mixers for each MAM lane, build a centralized paint production hub that supplies all lanes via a fluid bus. This reduces the number of Mixers and Crystalizers you need to manage and makes color supply problems easier to diagnose — if one lane is missing paint, the problem is in the lane, not in the paint source.
• Use separate control channels for each MAM instance. When you copy-paste a MAM to scale production, change the control panel settings, channel identifiers, and receiver configuration immediately. Two MAMs using the same control channel will produce the same shape and double-deliver to the Hub, wasting production capacity. Each MAM instance needs its own independent signal identity.

Frequently Asked Questions