The Signature House: The Next Great Firm Sells a Promise

In 1937 a young economist named Ronald Coase asked a question nobody had bothered to ask: if markets are so efficient, why do firms exist at all? Why not hire every task on the open market, one contract at a time?

His answer was that using the market is not free. Finding the right people, negotiating, writing contracts, checking the work, arguing when it goes wrong: all of that costs something. A firm exists to take those costs inside its own walls. And the walls sit exactly where it becomes cheaper to do something in-house than to arrange it with strangers.

That is why the four-firm arrangement exists. For a century, doing the work was the expensive, specialised part, and so was lending, insuring and lawyering. Each was hard enough to deserve its own firm, and the cost of the gaps between them was small by comparison.

Now change one input. Manufactured intelligence makes the doing cheap. The engineering, the drafting, the analysis all move toward the price of electricity. What does not get cheaper is the gap: the finger-pointing, the exclusions, the liability caps, the Monday morning when nobody pays. As the work shrinks, the gaps become the most expensive thing in the whole arrangement.

Coase tells you what happens next. When the cost of arranging something between separate firms is bigger than the cost of doing it under one roof, the walls move. The four firms fold into one. That firm is the signature house.

We have seen this move before. Construction used to split design and building between an architect and a contractor, and when something went wrong each blamed the other. The answer was design-build: one firm, one contract, one party responsible for the result. Owners paid for it gladly, because what they were really buying was the end of the finger-pointing.

A signature house does not sell hours, seats or licences. It sells an outcome with a guarantee attached. Every engagement has three parts, and each maps to one of the old firms.

This will be done

The engineering shop, running on manufactured intelligence, does the work.

To this standard

The legal side turns “good enough” into a precise, measurable promise that both sides can check.

And here is who pays if it is not

The insurer inside the house has priced the chance of failure, and the bank inside the house holds the capital to pay when it happens.

That is why it will look so strange to Silicon Valley. The Valley’s great invention was the product that ships “as is,” with every warranty disclaimed, sold at almost no marginal cost. The signature house is the opposite of a disclaimer. It carries reserves. It has to hold capital against its own promises, so it looks less like a software company and more like a bank that happens to employ engineers. Its most valuable asset is not code. It is its record of what went wrong and what it paid, because that record is how it knows what to charge, and nobody can copy it.

It will be born in the financial capitals for the reason I gave in the last piece: that is where capital, insurance and law already sit in the same few blocks, and where the financial sector argues its case to government.

1. The first signature houses are assembled, not founded.

   They will come from mergers across the old walls: an insurer buying an AI services firm, or an AI services firm standing up its own insurance company to back its promises. Watch for the first deal that puts a licensed insurer and an AI delivery shop under one roof.

2. Customers stop paying up front for AI work.

   When the doing is cheap, paying in advance for it makes less and less sense. Payment on acceptance, after the work is delivered, inspected and in use, becomes the normal way to buy AI services. Firms that insist on being paid before they stand behind anything will find it harder to win.

3. The rating becomes the brand.

   Customers will choose a signature house the way they choose an insurer: by its financial strength and its record of paying. A strength rating, not a feature list, becomes the headline in every proposal.

4. Regulators have to invent a category.

   A firm that does the work, insures it and finances it does not fit any existing licence. Expect a fight over whether it is an insurer, a contractor or a bank, and eventually a new class of licence with its own capital rules.

5. Every town gets small ones.

   The same shape works at local scale: a service provider that does the work, is paid only after the customer accepts it, and is backed by a local insurance agent and a local bank. The small signature house will be one of the best businesses a local operator can build.

Charleston AI is a service provider, and it makes and keeps promises in an order most businesses would call backwards. We do the work. The customer receives it, inspects it, puts it to use, and only then pays. The customer pays after.

That single choice changes the whole risk dynamic. Until the customer is satisfied, the risk that the work is wrong sits with us, not with them. There is no Monday morning where they call four firms and nobody pays, because they have not paid yet. In the language of this piece, it is a signature house in its simplest form: the work and the promise in one place, and the risk carried by the one who did the work. What a full signature house adds is the capital to make that promise for years instead of days, and for a hospital system instead of a restaurant.

This piece belongs to a set: When Intelligence Comes Off an Assembly Line, The Signature Is the Product, and Talent Will Gather Where Risk Is Priced.

Already true

Design-build already merged two professions into one point of responsibility. Specialist insurance for AI errors exists. Insurers are writing AI out of ordinary policies, which leaves more and more failures sitting in the gaps between firms. And pay-after-acceptance service providers already exist at small scale.

What has to happen

The gaps have to become more expensive than the walls. If AI failures keep landing in the seams, excluded by the insurer, disclaimed by the vendor, capped by the contract, customers will pay a premium for the one firm that closes them, and the merger logic takes over.

Where I am probably wrong

Regulation may keep the walls up. Insurers are separated from other businesses for good reasons, and a firm that insures its own work has an obvious conflict of interest. If regulators refuse to let the pieces combine, the signature house arrives as a tight consortium of separate firms bound by contract rather than a single company. The customer might barely notice the difference. The economics would be the same; only the org chart would differ.