A slicer for 3D concrete printingConcrete Labs2025 – 2026

What if slicing concrete were easier than slicing plastic?

The problem

Every concrete printer in the country waits for one person. The one who knows Grasshopper.

A machine that prints a wall in an hour was waiting a day for a file.

Students could model. They could not print. Operators could print. They could not slice. Between the two stood a definition only a specialist could open.

Where it starts now. Drop an STL. Works offline. Nobody asks who you are. — **Where it starts now.** Drop an STL. Works offline. Nobody asks who you are.

The reframe

Tools decide who gets to make things. So we asked a different question. Not how to slice concrete, who gets to.

The bar was Cura. Drop a file. See it in layers. Print.

If a fifteen-year-old can slice for a desktop printer, an operator should be able to slice for a six-axis arm. Four steps. In the order a whiteboard would put them.

Upload

The wall. Optionally a column and a reference. STL or Rhino.

Setup

Nozzle. Layer. Lean. Flow. The operator's numbers.

Infills & Seam

Pick rectangles on the plan. Pick where every layer begins.

Print Preview

Mass, volume, path, layers. Scrub it. Download.

A G-code answers every question at once. That was the problem.

Form, bead, seam, infill, machine, all frozen into one file by whoever wrote the definition. Untangle them and the product falls out.

The whiteboard. The architect ships geometry. The operator sets the bead. The machine holds the profile. Three people, three columns, one print file, assembled at the machine, not at a desk.

The nozzle belongs to the operator.

The architect draws the wall without knowing what will print it. Monday, a 25 mm nozzle. Tuesday, a 40. Same file. Nobody phones anybody.

As it arrived, and as it will print. Changed at the machine. Nothing upstream is touched. — **As it arrived, and as it will print.** Changed at the machine. Nothing upstream is touched.

One drawing, two beads. The outer line is the architect's. The offset is the operator's.

Infill is a site decision.

In Grasshopper, infill stops short of the back wall, runs through a column, leaves no room for one. Each is a call to the architect and a day lost. Here it is a rectangle picked on the plan by the person who can see the site.

Three ways a definition fails on a real wall, and the one that doesn't.

Rect X. Rect Y. Pick seam. The whole infill interface an operator needs. — **Rect X. Rect Y. Pick seam.** The whole infill interface an operator needs.

Plan. The outline the architect drew; the seam the operator chose. — **Plan.** The outline the architect drew; the seam the operator chose.

A seam you can see from the road.

Every layer starts somewhere. Left to chance it wanders and the wall carries a scar. Pick one point and every layer aligns to it, a line, not a scatter. Or spiral through it and never stop.

95 contours. Seam at 0°. Continuity OK. 761 corners kept sharper than 22°: the number a beginner would never know to set. — **95 contours.** Seam at 0°. Continuity OK. 761 corners kept sharper than 22°, the number a beginner would never know to set.

Delivery

Nothing to install. Nothing to license. Nothing to be down.

One HTML file. Built in the cloud, in a browser tab, in weeks, and it needs none of that to run. Ninety-five contours in about a minute, on the laptop in front of you. Nothing leaves the room.

Generating contours. On the operator's machine, offline. — **Generating contours.** On the operator's machine, offline.

Before a drop is pumped, the operator already knows.

The mass. The volume. The path. The layer count. A planter that was a set of sliders twenty minutes earlier.

layers

222m

path

147kg

concrete

0.067m³

volume

60,327

print points

Ready to print. The bead rendered at its real 25 × 10 mm. — **Ready to print.** The bead rendered at its real 25 × 10 mm.

The file. One line per point, layer breaks marked. Download. Walk to the machine. — **The file.** One line per point, layer breaks marked. Download. Walk to the machine.

One wall today. A building tomorrow.

Projects hold walls as assemblies, walls, columns, references, saved as a file the site can pass around.

Building Project 1. Designed in from day one, so the tool grows from a planter to a storey. — **Building Project 1.** Designed in from day one, so the tool grows from a planter to a storey.

The point

Concrete printing was never slow. Getting permission was.

The decisions underneath.

01 · Ownership

The nozzle is set by the operator, never by the file.

Form and bead were separated on the whiteboard before a line of code.

02 · Site

Infill is picked on the plan, not coded in a definition.

The person who can see the site decides what the site needs.

03 · Defaults

Printable by default.

35° lean, corners sharper than 22°, 0.6 mm tolerance, set, so a beginner never has to know.

04 · Seam

The seam is a choice.

Pick it, align to it, or spiral through it. A scar becomes a line.

05 · Delivery

A file, not a service.

Offline, on the site laptop. Nothing to install, license or be down.

06 · Scope

One wall, then a building.

Projects and assemblies were drawn in from the start.

Fact block

Product: Concrete Slicer
Inputs: STL · 3DM, wall, column, reference
Outputs: Print points (XYZ) / G-code text · project files
Print styles: Planar · non-planar · spiral continuous
Operator controls: Nozzle 4 – 80 mm · layer 2 – 50 mm · lean 0 – 55° · flow · tolerance · corner threshold · seam · infill rectangles
Machine profile: KUKA six-axis arm · concrete extruder
Form: Single HTML file · three.js · Clipper · runs offline
Test print: 950 mm vessel: 95 layers · 222 m · 147 kg · 0.067 m³
Years: 2025 – 2026
Role: Product strategy · interaction design · build