Kirkland Is Building Its Own Kitchen. Most Law Firms Just Need to Learn to Cook - on law firm AI strategy

TL;DR  Kirkland & Ellis is spending $500 million to build its own AI, on the logic that off-the-shelf tools are “raising the floor for everyone” and that the firm does not get hired for the floor. Both halves of that are true. They just lead to the question Kirkland never answered: if you don’t get hired for the floor, why build the floor yourself? No restaurant earned a Michelin star by building its own oven. For almost every firm that is not Kirkland, the move is to buy the best oven everyone can buy, and win on the menu. Interested in more? join the webinar here: https://luma.com/7mk7c7xe

Last week (late May 2026), the Financial Times reported that Kirkland & Ellis, the highest-grossing law firm in the world, had set aside $500 million to build its own artificial intelligence platform from scratch, built by its own engineers, trained on the workflows of 250 of its lawyers, owned outright and sold to no one.

The firm’s chair, Jon Ballis, gave the quote that explains the whole decision. Widely available AI tools, he said, are “raising the floor for everyone” in the legal industry. But Kirkland had to do more than use them, because “we don’t get hired for the floor.”

He is right. On both counts. Which is exactly what makes Kirkland’s conclusion so strange.

If you don’t get hired for the floor, why build the floor yourself?

That is the question the $500 million never answers. And most of the profession, hearing only the headline number, drew the opposite lesson from the one Ballis’s own sentence supports.

You don’t build your own oven

The answer is sitting in the best restaurant you have ever eaten in.

No three-star restaurant earned its stars by building its own oven.

The oven is the floor. Every serious kitchen has a great one. They buy the best oven money can buy, the same oven their rivals can buy, and then they go and win on everything the oven cannot do for them: the ingredients, the recipes, the craftsmanship, the service, the room. The oven is table stakes. Nobody books a year ahead for the oven.

Ballis is correct that the floor is rising and correct that you are not hired for the floor. His error is in the very next step: deciding that the response is to build the floor yourself. To forge your own oven.

At Kirkland’s scale, that flex is defensible. The firm crossed $10 billion in revenue last year. When you are that large, building your own kitchen is a rounding error, and owning a piece of infrastructure no competitor can licence is a real, if expensive, moat. Fine.

Kirkland can do what Kirkland is doing precisely because it is Kirkland.

The danger is everyone else watching this and concluding they are behind because they have not started forging an oven of their own. Because they are not behind. And if they copy the Kirkland behaviour, they are about to make the single most expensive mistake available to them: spending scarce capital to build something a supplier already sells, instead of spending their scarce attention on the thing clients actually pay for.

Confusing the floor with the kitchen

The build-versus-buy question only looks hard if you confuse the floor with the menu.

For a firm with a hundred lawyers or fewer, building proprietary AI is not a moat. It is a way to go bankrupt proving you can do what a vendor does better, cheaper, and faster than you ever will. The minimum cost of building is enormous and mostly fixed: engineers, infrastructure, maintenance, the constant race to keep pace with models that improve every few months. A giant amortises that across thousands of lawyers and billions in revenue. You cannot.

One percent of revenue buys Kirkland a platform and an engineering team. At a mid-sized firm, the same one percent buys some software licences and a part-time lead.

So the first move, for everyone who is not Kirkland, is almost embarrassingly simple. Buy the oven that best suits your practice. The one everyone can buy. And stop feeling behind for not building it. The floor is supposed to be bought. That is what makes it the floor. It is dull advice, the kind nobody writes a profile about. It is also the right advice.

Which leaves the only question that was ever interesting.

So what is your menu?

If the oven is settled, the whole game is the menu.

This is where strategy lives, and none of it comes in a box. Your sourcing is the clients and the matters you choose to obsess over. Your craftsmanship is the judgment your best lawyers spent decades building, the kind that does not transfer in a software licence.

Your signature dish is the one problem you decide to solve better than anyone, the reason a client picks you over the firm down the road running the identical oven.

None of that is bought. All of it is chosen. So ask yourself: if the firm across town bought the exact same tools tomorrow, what would still make a client choose you? That answer is your menu. If you cannot name it, a faster oven will not save you. It just gets you to mediocrity quicker.

So, again, for the people in the back: the tool itself does not give you a menu. It will happily serve a faster, cheaper version of the same food as everyone else. If you want to stand out: focus on your menu instead.

What a law firm should actually do - on law firm AI strategy

So what should be your law firm's AI strategy? Stripped of metaphor, the playbook is short.

Buy the floor, like everyone else, and refuse to feel behind for it. Increase productivity in the short term, and use the increased efficiency it gives you to fund one transformation. Then choose your signature dish: the single client type, workflow, or outcome you will be known for, and the one deliberate thing you will do with these tools that the firm down the road will not. Run that one bet with discipline, protected from the day-to-day machine that will otherwise strangle it. And adapt your pricing accordingly, because a kitchen that gets dramatically more efficient and keeps billing by the hour is quietly punishing itself for getting better.

That is the whole thing. Buy the oven. Choose the menu. Cook one dish better than anyone. If the floor is not what you are hired for, it is not what you should be building.

Buy it. Then go cook.

Interested in more? join the webinar here: https://luma.com/7mk7c7xe