The legal business runs on a meter: time billed. That model has survived everything for a century. Now a technology shows up that takes specific tasks — first-draft documents, summarizing a stack of discovery, the rote research pass — and collapses an hour of them into a couple of minutes.

If your revenue is hours, and a machine just deleted a pile of hours, that reads like a threat. I don't think it is. I think it's a forcing function that finally separates the hours clients were happy to pay for from the ones they always quietly resented. I build AI for small businesses, I'm not a lawyer, and that outside view is exactly why I'll say the part the profession is tiptoeing around.

The hours AI eats were mostly the hours clients hated paying for anyway. What's left — judgment, strategy, trust — was always the real product. The model just made that impossible to hide.

Two kinds of hours

Every firm bills two completely different things under the same word, "time."

There's leverage time — drafting the standard motion, reading two hundred pages to find the five that matter, the research any competent associate would do. The client pays for it, but they don't value it; they value the outcome it produces. They'd pay zero for the hour if they could get the result without it.

Then there's judgment time — what to actually do, how to read this judge, whether to settle, the read on a situation that comes from having seen a hundred like it. That's not a task. That's the thing they're actually hiring.

AI comes for the first kind, hard. It barely touches the second. The firms that panic are the ones who've been quietly making most of their margin on leverage time and calling it expertise.

This already happened to me, in reverse

I watch lawyers brace for this and I recognize it, because I came up in construction, where the same split exists. There's swinging-the-hammer time and there's knowing-what-to-build time. The trade taught me early that the hammer hours are the commodity and the judgment hours are the business.

When I started building software, AI ate a giant share of my "hammer hours" — boilerplate, lookups, the grunt work. It didn't make me less valuable. It made the grunt work cheap and pushed all my value into the part that was always the point: deciding what to build and making sure it's right. The meter didn't kill the work. It moved the work to where the work always lived.

What a smart firm does about it

The move isn't to ban the tools to protect the meter. Clients will find out, and "we bill you for hours a machine could do in seconds" is not a pitch that survives the next decade. The move is to let AI take the leverage hours and reprice around judgment:

  • Hand the rote tasks to the machine — intake, first drafts, discovery summaries, research passes — and stop pretending those hours were the product.
  • Pass some of that speed to the client. Faster and cheaper on the commodity work is a moat, not a loss, because the firm down the street is still billing four hours for it.
  • Charge for outcomes and judgment, the things that were never really an "hour" in the first place.
  • Own the tooling. If AI is going to handle client matters, that system and its data belong inside the firm, not rented from a vendor who's also serving the firm across town.

The honest bottom line

A technology that collapses billable hours is only a threat if your value was the hours. If your value was judgment, AI just stripped away the busywork that was diluting it and let you charge for the real thing.

The firms that win the next ten years won't be the ones with the most associates grinding leverage hours. They'll be the ones who let the machine do that, kept the data in the building, and got honest that judgment — not time — was always what the client was buying. The smart place to start is small and safe: the front door, not the courtroom.