The Master Plan, Part Two

In March I wrote the master plan: make it joyful to own real estate, in three steps. Predictive maintenance first, then properties that can act on their own behalf, then buildings that improve themselves.

Eight months later the destination is the same. What changed is the shape. We thought we were building a product that sits on one side of a building. We're building the network that sits between all of its sides, and the three steps are re-cut around that.

Every transaction has two sides

Look at anything that happens to a building and you'll find a pair.

• A buyer and a seller.
• A lender and a borrower.
• An insured and an insurer.
• A landlord and a tenant.
• A building owner and a roofing company.
• A facility and the crew that services it.
• A data center operator and the contractor who closed up the ceiling.
• A city and whoever is asking it for a permit.

Each pair makes its own record of the same building. The appraisal doesn't talk to the insurance inspection. The roofer's invoice never reaches the lender's collateral file. The drawings of a data hall stop being true the day the containment goes in, and nobody on either side of the maintenance contract has a current one.

The building is the one thing every pair has in common, and it's the one thing none of them share a model of. That's the gap.

What Visa figured out

The Acquired episode on Visa is the clearest explanation I know of what a network is.

In 1958 Bank of America mailed sixty thousand unsolicited credit cards to the people of Fresno, California, and the BankAmericard was born. It worked, badly. By the late sixties the program had been licensed to banks all over the country, each one running its own version, and the settlement between them was a mess of paper and bad debt.

Dee Hock, who ran the BankAmericard program at a bank in Seattle, convinced the licensees to do something strange: take the network away from Bank of America and give it to all of them. In 1970 the banks formed National BankAmericard Inc., a membership corporation owned by the banks that used it and controlled by none of them. In 1976 it became Visa.

The insight that made Visa one of the most valuable companies on Earth is that the valuable thing was never the bank that issued the card, or the merchant that took it. It was the thing in the middle: a shared set of rules and rails that let any two strangers transact and settle, owned by no single side, taking a sliver of every transaction that crossed it. Of the roughly 2% a merchant pays on a card payment, the network keeps something like 0.15 to 0.2%. On $14 trillion a year, that sliver is a half-trillion-dollar company. The hosts' line for it: a toll booth. Their other line: the NFL doesn't own the teams; the team owners own the NFL.

Issuers compete with each other. Merchants compete with each other. The network doesn't compete with anyone. It takes no financial risk and just moves information around. It only has to be true, neutral and everywhere.

The twin is the rails

The thing in the middle, for a building, is its digital twin. Not a PDF, a spreadsheet, or a photo album on someone's phone. A model of the building that every side can read and that gets more true over time.

So the company does three things, and only three.

Capture makes the twin. An iPhone Pro walk-through comes back as a 3D model of the space in minutes, and it gets sharper on every visit.

Edit keeps it true. Pascal Editor is open source, free, and built so that both people and agents can work on the same model, then export it to the formats the other sides already use.

Service writes reality back into it. A dedicated crew does the work, and every visit lands in the model as photos, notes and changes. A twin that nobody maintains is a drawing. A twin that gets serviced is a ledger.

Open source is not a marketing choice here. Nobody joins a network owned by their counterparty. Visa's members owned Visa. The editor belongs to everyone who uses it, which is the only way the lender, the insurer, the roofer and the owner will all trust the same model.

The plan, restated

Same three steps, re-cut around the network. Here's what they look like now.

Step one, now: capture, edit, service. This is predictive maintenance from the first plan, and it's how the twin gets created and kept true. For homes it's a membership, live in Florida: one dedicated team, every major system looked after before it becomes expensive, everything in one place. For commercial buildings it's a contract, starting with data centers, the buildings that most need a model that survives change. Every building on the platform is accumulating the context the next two steps run on.

Step two: sovereign properties. A building with its own financial identity. Every home owned by a trust, with Pascal as the rails for spinning one up, updating it, and transferring ownership in real time. No title company, no lawyers, no friction: buy a home in a day, like a phone or a car. The twin is what makes this safe. Both sides of the sale, and the lender behind one of them, are reading the same building.

Step three: the financial layer. An AI bank and an AI insurer purpose-built for buildings, where everything from mortgage origination to repair reimbursement is handled by agents reading the twin. Real-time scans price risk, detect fraud and schedule repairs. The building pays its own bills, books its own maintenance and applies for its own insurance. This is the first plan's self-improving building, with the money attached.

Long term. Every building on Earth represented as a digital object: scanned, insured, maintained, and legally owned by a trust. People shouldn't be landlords or accountants or project managers. They should live.

Why a network and not a product

Products scale with customers. Networks scale with pairs.

Every side we add makes the twin more valuable to every other side already on it. The roofer's visit makes the insurer's risk model better. The insurer's requirements make the owner's maintenance plan better. The lender's collateral file stops being a stale appraisal and becomes a live building. None of that happens if the model lives on one side.

It's also why the order matters. You can't give a building financial agency until you understand its physical condition in real time. You can't let it improve itself until it can pay for the work. Each step is funded by the one before it and makes the next one possible.

The market

The U.S. housing market is worth about $50 trillion. The friction cost of a transaction is roughly 10%. Homes change hands every seven years on average, so about 1.4% of that $50 trillion flows every year as fees, roughly $700 billion.

What's new is the other side of the ledger: the buildings that can't go down. The four largest cloud companies have guided to well over half a trillion dollars of capital spending in 2026, most of it data centers. North America has tens of gigawatts in the pipeline, and nearly all of what is under construction is already leased. And the people who run them are stretched: in Uptime Institute's 2026 survey, 53% of operators say they can't find qualified candidates, and nearly half had an outage that mattered in the last three years. Those buildings need a true model more than any building ever has. They're where commercial starts.

---

Same order, different shape, one mission: preparing the physical world for AI. If you want your building on the network, start with a scan, or talk to us about a commercial building.