I'm a big fan of AI. I think it's genuinely transformative—for the practice of law and, as I've written elsewhere, for access to justice. It's been a pleasure to write about, and lately I've had the chance to present to other lawyers on integrating AI into their legal practices. So I read this Law360 piece on how firms are approaching AI adoption with real interest.

One line stopped me. Larry Gresser of Cohen & Gresser explained why his firm didn't rush to buy the flashiest platform: they were focused on "building the road before buying the Ferrari"—getting their systems, processes, and data ready first.

That's exactly right.

It reminds me of law school. The exam wasn't really about the three hours of writing. It was about the outline—the weeks of synthesizing cases and building a framework you could use under pressure. Show up with a great outline, and the exam runs smoothly. Show up without one, and no amount of raw ability saves you.

AI tools are the same. The model is powerful, but it performs only as well as its foundation: clean data, sound processes, sharp prompts, and lawyers who know the law well enough to direct and check the output.

So build the outline. Then take the exam.