Intent Modeling vs Tectonic Modeling

OnlyPlans: Intent-Based Design vs. Tectonic Modelling

(a.k.a. Why I’m Done Drawing the Same Wall Twice)

Ever finish a 60-hour week and wonder why you’re redrawing that same damn wall section for the fifth project in a row? Yeah, me too.

Let’s talk about two very different ways to model buildings—and why one of them scales, and the other one just burns time. Here’s the whole difference in one line: a tectonic model stores what you drew; an intent model stores why you drew it. Only the second one knows what to redo when something changes.

Intent-Based Design: Let the Computer Do the Boring Stuff

This is the world I live in.

You start a design by telling the system what it is: “Multifamily over podium, 1B zoning, 62% efficiency, 5 stories, 1-hour rated corridors.” (Efficiency here is net usable area divided by gross area.) That’s your intent. The machine handles the rest. It figures out the layout, the fire rating logic, the unit counts, the stairs, the trash chute. It’s not magic—it’s just logic that’s been encoded.

Intent modeling, in plain terms: you model what each element has to do and what it depends on—the two rooms a wall sits between, the line a cabinet hangs from, the rating a corridor needs—and the machine derives the geometry and the assembly from that. The model keeps the reason. The lines are output.

The goal isn’t to replace the architect—it’s to elevate them. When the system handles the low-leverage stuff, you get to focus on the real work: context, form, flow, experience. Design.

It’s not about less thinking. It’s about thinking better.

Upside: Your decisions scale. You can explore more options, faster.

Downside: You’ve gotta have systems in place. And yeah, some of them will suck until you fix them.

Tectonic Modelling: Manual Mode

This is how most of us were trained. You start with a blank wall. Add the studs. Add the gyp. Insulation. Vapor barrier. Sealant. Every layer, a click. Every detail, a choice.

Tectonic modeling takes its name from tectonics, the architect’s word for the craft of how a building is physically put together. You model the thing as built: each layer placed, each dimension typed. The model keeps the result. The reasons stay in your head.

It’s got a certain romance to it. You’re crafting something. But let’s be real—it’s slow. And if you’re doing production work, it’s wasteful. That same wall you built on Project A? You’re rebuilding it on Project B, even though, by my guess, 98% of it is identical.

Tectonic modelling gives you ultimate control. But it doesn’t scale. It doesn’t teach the machine anything. It doesn’t create leverage.

Upside: Pure control. Every line is yours.

Downside: Every mistake is too.

One Wall, Modeled Both Ways

Take two apartments side by side. Unit A’s living room meets Unit B’s bedroom along a 30-foot wall. Somebody drew a plain partition there, call it 5 inches thick. Then somebody notices it’s a unit separation, and the rated, acoustically separated assembly comes in at about 10 inches. (Round numbers for the arithmetic, borrowed from Walls Between Cells, not a spec.)

In a tectonic model, that wall is a 5-inch object at a set of coordinates. Nothing in the file says it sits between two dwelling units, so nothing in the file knows it’s now wrong. Somebody has to notice, delete it, draw the 10-inch type, and decide—usually by accident—which side it grows into. Next project, same wall, somebody has to notice again.

In an intent model, the wall isn’t an object first. It’s a pair: Unit A on one side, Unit B on the other, an edge between two rooms. (That’s the whole argument of Walls as Edges.) The pair says which assembly the wall needs, and a datum rule—which line holds still when the wall gets thicker—says where the extra 5 inches go. Hold the centerline and each unit gives up 2.5/12 × 30 = 6.25 ft². Hold Unit A’s face and Unit B’s bedroom gives up 5/12 × 30 = 12.5 ft². The intent model doesn’t pick for you. It makes you pick once, on purpose, instead of letting whoever dimensions the last cabinet pick on every project.

That downside I mentioned is real, though. In Elevation I I rebuilt a kitchen elevation as a host graph, where every object hangs from a parent: the floor, the wall, the ceiling or another object. The upper cabinets hang from the ceiling. Raise a 120-inch ceiling to 132 and the model obediently lifts the bottoms of the 30-inch uppers from 84 to 96 inches, and the gap over the 36-inch counter goes from 48 inches to 60. Five feet of air over the counter. A tectonic model would have left those cabinets at 84—not because it knew better, but because it didn’t know anything. The fix isn’t to go back to typing 84. It’s to fix the parent: decide what the uppers really hang from, so every kitchen after this one inherits the answer.

So What Do We Do With That?

We blend it.

We use intent-based systems to get to 80%—fast. Then we layer in tectonics where it matters. On the joints, the transitions, the moments that make the building sing. But we don’t start from scratch unless we have to.

The real play is in shortening feedback loops. That’s how you unlock iteration. That’s how you stop saying “it’s too late to change” in week 10 of design.

When the boring stuff’s automated, we get to spend our time solving better problems—like housing, affordability, livability. The stuff that actually matters.

Bottom Line

Tectonic modelling is for when you don’t trust your system yet.

Intent-based design is for when you do.

Here’s my test for any element. If you can name what it depends on—the two rooms a wall separates, the line a cabinet hangs from—model the dependency and let the geometry fall out. Draw layers by hand only where the joint itself is the design. And when a rule gives the wrong answer, fix the rule, not the drawing. A fixed drawing helps one project. A fixed rule helps every project after it.

So build your systems. Teach the machine. And let it take care of the repetitive work—so we can get back to doing what we’re actually good at: designing.

Until next time, stop drawing the parking stalls. Start fixing the housing crisis.