Who Gets to Build? Vibe Building for the Real World

I own a piece of land in Berkeley. Last year, I started trying to build on it. I quickly understood why most people don't.

I'm not unusual. Millions of Americans own land they could build on — backyards in expensive cities where the zoning now allows an ADU, rural parcels that could host a cabin, urban infill lots waiting for someone to break ground. Most of that land sits empty.

Not because the owners don't want to build. Because they can't.

The bottleneck isn't capital. It isn't dirt. It's expertise.

Between a person who owns land and a person who can swing a hammer, there's a gap most people never cross: zoning lookups, setback rules, soils reports, structural calculations, energy compliance, plan sheets, permit submittals, redlines, resubmittals. The vocabulary alone takes months to learn. The actual work takes a small team of licensed professionals who, in most U.S. cities, charge between $20,000 and $80,000 to get you to the point where a city will let you put a shovel in the ground.

That's the expertise gap. And it's the reason most land doesn't get built on, even when the country desperately needs more housing.

What we're building

Terraformal turns a description into a permit-ready building.

You tell it, in plain language, what you want to build — a 600 sq ft ADU in your backyard, a 1,200 sq ft cabin on a rural lot, a small commercial space on Main Street. Terraformal walks you through it: clarifying questions about your land and your priorities, surfacing what fits within your jurisdiction's code, refining the design through iteration, matching to the right pre-approved templates where they exist, generating the documentation a city actually wants to see.

Think of it the way Cursor approaches code. You don't have to know everything about syntax, libraries, or build tools to make a working app — you describe what you want, and Cursor pairs with you to get there. Terraformal is the same idea, applied to the much harder real-world version: an iterative AI partner for the construction stack.

We call it Vibe Building for the Real World.

Describe it. Shape it. Build it. Permit-ready.

Why now

Three things are converging.

AI got good enough. The last eighteen months changed what software can reason about — not just text completion, but the structured, jurisdiction-specific, regulation-laden domain of building codes. A model that can read a zoning ordinance, cross-reference a property's parcel data, and translate that into a buildable footprint wasn't possible two years ago. It is now.

The regulatory ground is shifting. California's AB 1332 now requires cities to publish pre-approved ADU plans online. Statewide ADU rules have been pried open. Cities across the country are streamlining the permit process for small residential structures because the housing crisis demands it. The plumbing for democratized building is being laid. What's missing is the interface.

The people who'd benefit most are already here. Homeowners with backyard lots. Landowners with parcels they've inherited. First-generation families stretching every dollar to build housing for the people they love. Small developers who can't justify a full architectural team for a single ADU. None of them are served well by the existing professional stack. All of them could be served by software that meets them where they are.

Who gets to build

The question we keep coming back to is the title of this post.

For most of the last century, the answer has been: people who can afford a team of specialists. Architects. Engineers. Permit consultants. Contractors who know which forms get filed in which order. The gatekeepers aren't malicious — most of them are extraordinary at their jobs, and the most ambitious projects will always need them. But the cumulative effect of needing all of them, in sequence, to build anything is that an enormous number of people who could be building, aren't.

The answer isn't to remove professionals from the loop. Architects and engineers will keep doing the work that requires their judgment, and Terraformal will route that work to them when it's needed. The point is that the first mile — the part where someone with a lot and an idea figures out whether what they want is even possible, and what shape it could take — should not be locked behind a $20,000 retainer.

If we get this right, the answer to "who gets to build" expands. Not just to professionals. Not just to people with capital and connections. To anyone with land and an idea.

That's the version of the future we're building toward.

What's next

Terraformal is live at terraformal.ai. We're starting with ADUs in California — the highest-density combination of housing demand, regulatory tailwind, and pre-approved template availability anywhere in the country. From there, we'll expand outward: small homes, accessory commercial, and the broader long tail of small-scale construction that the existing stack ignores.

This blog will follow the work. Some posts will be technical — how we ingest jurisdiction-specific code, how we model permit workflows, what happens the first time AI gets a setback wrong, and we have to figure out why. Some will be about specific projects we help get built. Some will be about the broader question of what software can and can't do in the physical world.

If any of this resonates — whether you're a homeowner with a lot, a builder, an architect, a city official thinking about how to streamline permits, or just someone who cares about the housing question — we'd love to hear from you.

The land is already there. The hammer is already there.

It's time to close the gap in between.

— Nicolás