Bespoke Material Systems

For most of design history, you designed the thing - a chair, a panel, a beam - as a singular object. Digital fabrication invites a different move: design a component and a rule for combining components, and let the whole emerge from their aggregation. This is thinking in systems, and it is how a huge amount of contemporary fabrication-led architecture and interiors actually work. A perforated screen is not drawn hole by hole; it is one tile plus a rule that varies the perforation by daylight. A gridshell is not modelled member by member; it is one node-and-strut relationship, repeated across a surface.

The shift matters because it changes what you control. When you design a system, you set the DNA - the unit, the joint, the variation rule - and the specific instances follow. You gain reach (thousands of coordinated parts from one definition) and coherence (everything shares a logic), and you trade away the illusion of placing every piece by hand. The interest, and the design intelligence, moves into the rule: how the unit varies, how it connects, how it adds up. A weak system is a grid of identical tiles; a strong one is a unit and a rule rich enough that the whole feels inevitable and alive.

There is a real economy to this too. Fabricating a thousand bespoke parts by drawing each one is impossible; fabricating them from one parametric definition is routine, because - as Module 0 argued - the machine charges the same to read a different file. Systems thinking is therefore not just an aesthetic; it is how you make ambitious, high-resolution work affordably. You invest effort once, in the DNA, and amortise it across every instance the rule produces.

A material system starts with a unit - the repeating component that will be fabricated many times. The craft is choosing a unit that is cheap to make in quantity yet capable of variation. Because digital fabrication makes each part from its own file at no extra setup cost, the unit does not have to be identical every time; it can flex within limits - a tile whose fold angle changes, a strut whose length varies, a panel whose opening scales with sun exposure. This is controlled variation: same family, same fabrication logic, tuned parameters.

Getting the unit right is mostly about restraint and manufacturability. Keep the fabrication simple - if each unit is a nightmare to cut, a thousand of them is a thousand nightmares. Keep the variation parametric, so every instance comes from one definition rather than a thousand hand-edits (Module 9 makes this pipeline explicit). And keep tolerance in mind at system scale: a 0.2 mm error per unit is invisible in one tile but stacks into a 20 mm gap across a hundred - so systems need either self-registering joints that absorb error or a deliberate tolerance strategy. Design the unit as a species, not a single specimen: define what stays constant and what is allowed to change, and you have the genome of the whole system. A good test is to ask what a fabricator would need to make any instance: if the answer is one cut file and a handful of parameters, the unit is well-defined; if it is a folder of bespoke drawings, it is not really a system yet.

If the unit is the vocabulary, the connection is the grammar - and it often matters more. How repeating parts meet decides how the system is cut, assembled, aligned, and whether it can ever come apart. Three broad strategies recur. Integral joints - tab-and-slot, finger joints, notches cut into the parts themselves - let components self-register: the geometry aligns the pieces, so assembly is fast and forgiving and no separate fasteners are needed. This is the darling of laser-cut and CNC systems because the joint is made in the same cut as the part. Shared nodes - a common plate, hub or connector that many parts bolt or slot into - suit systems where members meet at varied angles, like space frames and gridshells; the node absorbs the geometric complexity so the struts can stay simple. Interlocking or stacking - parts that hook, nest or stack without fasteners - give reversible, tool-free assemblies.

The choice ripples through everything. Integral joints are elegant and cheap but lock the assembly sequence and can be fiddly to take apart. Shared nodes cost more parts but handle irregular geometry and make disassembly easy - which, as the sustainability lesson argues, is a real virtue. Choose the connection logic early, because it constrains the unit, the fabrication and the whole life of the system. The joint is not a detail you add at the end; it is a primary design decision.

The third ingredient is the rule - the logic that generates the variants and arranges them into the whole. In practice this is where parametric and computational design meet fabrication: a definition (often in Grasshopper or similar) takes a base surface or field, tessellates it into units, varies each unit by some driver, checks that every instance stays fabricable, and outputs the cut files. The rule might tessellate a curved surface into flat panels (planarisation, so each panel can be cut from sheet), vary a screen opening by incident sunlight, or size structural members by the load they carry.

A good rule does three jobs at once: it generates the geometry, it rationalises it so every part can actually be made (flat where it must be flat, within machine size, within tolerance), and it documents it - numbering parts, nesting them on sheets, producing an assembly key. This is the honest, unglamorous backbone of a real system: without rationalisation, an ambitious surface yields a thousand impossible-to-cut parts; without documentation, a thousand near-identical pieces become impossible to sort and assemble on site. The rule is where designers spend most of their effort, and where a system succeeds or collapses. Ambition is easy; a rule that keeps ambition buildable is the skill.

Look around and material systems are everywhere once you see them. A perforated or folded metal facade is one panel and a perforation rule. A timber gridshell is one node-and-lath relationship repeated over a doubly-curved surface. A laser-cut acoustic screen is one tile whose depth varies with the sound field. Research groups have made this a whole practice - Stuttgart’s ICD builds pavilions from thousands of individually-fabricated timber or fibre components governed by a single computational system, and Gramazio Kohler assemble brick and timber systems robotically, each unit placed by rule. The common thread is never the individual part; it is the unit-connection-rule triad producing a coherent, high-resolution whole.

When you next meet an impressive fabricated surface, reverse-engineer it: what is the repeating unit, how does it connect, and what rule varies and arranges it? That reading is the same skill as designing one, run backwards. And it comes with a discipline: a system is only as good as its buildability and its afterlife. Design the connection so it can be assembled by real people in a real sequence, and ideally disassembled and reused - because a bespoke material system that cannot come apart is a bespoke pile of waste. Systems thinking and sustainability, the next lesson, are two sides of the same coin.