Legora, a legal AI company now worth $5.6 billion, published a report on the return its customers get from the tool. The research, by legal analyst Ari Kaplan, covers customers at 31 firms, so read the numbers as directional, not gospel. It reads like a victory lap for efficiency. What it records is a small group of firms changing what they sell.
Start with what’s true. The time savings are real. Acquisition reviews that took weeks now take hours. Document review that took months takes days. I’m not going to pretend that isn’t valuable. It is. Efficiency is the on-ramp. The mistake is treating it as the destination.
The hours AI frees up have to go somewhere. They come off the client’s invoice, they go back into more of the same work, or they pay for something new the firm can sell.
The savings have to land somewhere
Some firms in the report let them leak straight to the client, as the same work delivered faster at a smaller bill. Some poured the freed hours back into more of the same work, and the vendor takes its cut of that one: a bigger seat, more of the firm’s people on it, a deeper hold on the workflow the firm now runs on. A smaller group put the time into something they couldn’t sell before: 42% said the tool helped them win new work, and many said it made fixed and capped fees easier to offer. That group is selling outcomes instead of hours, and it has an advantage its competitors can’t buy next quarter.
The client never saw it
The wins in the report are incidental. New work arrived “directly or indirectly.” The report ends by admitting most firms can’t measure any of it yet. They backed into commercialization. They didn’t design it.
In the same months a firm logs an incidental new matter, its clients report no change in the work, the price, or the speed. The value moved and nobody built it to be seen. That’s the difference between a good quarter and a new line of business.
What you rent, they rent too
This is why efficiency alone caps out. The capability arrives in a seat, and your competitors buy the same seat the same week. Whatever edge it gives you becomes table stakes, and after that the only lever left is price.
The vendor isn’t standing still either. Legora spent this year buying four other companies, including one that sells commercial real estate software, a long way from “AI assistant for lawyers.” It’s buying domain depth as it goes. What you rent today, your client can rent next year.
What you can’t buy in a seat is your own judgment and domain expertise, packaged into an offering with a price on it. That’s the asset. Efficiency is what frees the time to build it.
The report ends where the commercial question starts. Time savings arrive on their own. The offering you sell with the freed time is the part a firm builds on purpose, and you’ve already done the margin math that sizes it. Hours nobody assigned to a new offering land in the client’s column, and the invoice shrinks by exactly what the tool saved.