In the AI era, your company will look like a law firm

Everyone says AI lets one person do a team's work. True — but I think most people draw the wrong conclusion from it.

Ten years ago, "being able to build the thing" was itself a moat. People who could write software — and actually finish it — were scarce. Shipping alone made you win.

AI has pushed the cost of building to the floor. Being able to build is no longer the moat. It's the entry ticket.

So where did the game move? To two things: whether you can spot a demand that actually exists, and whether you can stand behind the outcome — all the way to the end.

Here's the trouble: these two almost never live in the same person. People who understand demand usually aren't the ones who can do top-tier work. People who do top-tier work are usually far from the client and the market.

There's a hundred-year-old structure that solved exactly this: the law firm. On one side, partners — they hold the client, understand what the client actually needs, and answer for the outcome of the case. On the other side, associates — they do the casework. Demand and execution get split in two, each pushed to its best, then welded back together by one thing: accountability.

The more I think about it, the more I believe this is the right shape for an AI-era company. It's how we actually run today: I'm up front catching demand, talking to clients, owning outcomes; three architects behind me land the projects one by one and make them solid.

And one thing I'm increasingly sure of: technical service will standardize, the way legal service did. As coding agents get stronger, the "build it" end looks more and more like a repeatable assembly line. Who does the building matters less and less.

Which is exactly why the game moves upstream. The hard, valuable thing now is running the whole chain: from catching a real demand, to delivering it in an increasingly standard way, to answering for the result at the end. Getting that chain to run smoothly — that's the real fight ahead.

If you're thinking seriously about how a company or a team should be built in the AI era — I've been chewing on this for a while, and I'd genuinely like to compare notes. Write me: [e-mail]