Harvey's Law Firm Stories Reward the Firms That Prepared

On October 21, 2026, Harvey, a company that sells AI software to law firms, published a case study on its own blog about how mid-sized firms are using its tool. You can read Harvey's post on mid-sized firms and AI adoption for the full set of examples. The firm with the headline result, Munck Wilson Mandala, has nearly 100 lawyers. It's a vendor case study, so every figure below is Harvey's or the firm's own account, not an independent test.

Here's our honest take. The results are striking, and the firm reporting the biggest one had done real groundwork before it ever ran the project. If you lead a firm competing against bigger teams, our read is that the payoff came from getting ready first, and buying the tool came second.

One project at Munck Wilson Mandala

The headline example is a due diligence project. Due diligence is the document review a law firm does before a deal closes, checking contracts and records for risks the client needs to know about. It's usually done by associates, the salaried lawyers who aren't yet partners and do much of a firm's document work.

According to Harvey's post, Managing Partner Bill Munck said the project would have taken three associates working a week and a half. On Harvey, it was done in a day. Harvey reports the output achieved more than 90% accuracy, though we didn't see how accuracy was measured, so we'd treat it as the firm's and Harvey's figure.

That's one firm and one project, so we read it as an example rather than a typical result. Still, for a managing partner, a week and a half of work landing in a day is hard to ignore.

Competing against a bigger bench

In our read, large firms can have an edge on document-heavy work like due diligence. They can put more associates on a matter, which is a single client case or deal, and finish faster. Lawyers call the ratio of associates to partners associate leverage, and a firm with more of it can staff big projects quickly.

A firm of 100 lawyers, or smaller, competes against that with fewer hands. When a client wants diligence done on a tight deal timeline, team size can weigh on who gets the work. That's the pressure Munck's example speaks to.

If AI can take on part of the review, a smaller team may be able to keep pace on work that used to need a larger bench. That's our read of what the example suggests, not something Harvey's post proves across firms.

The groundwork that came before the result

Here's the part of Harvey's post we think matters most, and it's easy to skim past. Roughly half of Munck Wilson Mandala's nearly 100 lawyers hold STEM degrees, meaning science, technology, engineering, or math. The firm also had a multidisciplinary group evaluating AI before generative AI took off. Generative AI is the kind of AI that writes and summarizes text.

The firm also has an AI use policy. It requires lawyers to tell clients about AI use and get their permission. That's a rule written down before the work starts, not after.

We don't read the STEM degrees as something other firms must copy. The lesson, in our view, is the habit: a group that studied AI early, and a clear policy on how it's used with clients. Any firm can build those, whatever its lawyers studied.

What one Gray Reed attorney reports

Harvey's post also describes Tanner McClellan, a transactional attorney at Gray Reed. A transactional attorney handles deals and agreements rather than lawsuits. According to Harvey, McClellan saves approximately three hours a day using the tool.

He uses it for two jobs, per the post. One is updating hundreds of pages of documents after a deal's structure changes. The other is comparing lengthy credit agreements, the contracts that set the terms of a loan.

That's one named attorney's account, not a firm-wide average. We'd be careful about multiplying it across a whole staff. What it does show is where the time goes on deal work: long documents that need to change in many places at once.

Adoption numbers at Foley Hoag

The third firm in the post is Foley Hoag. Harvey reports that more than 94% of the firm's lawyers have adopted the tool. It also reports that over three-quarters of the firm's lawyers and staff use it weekly.

Those are two different measures, and it's worth keeping them apart. The first counts lawyers who adopted it. The second counts lawyers and staff together who use it every week.

Adoption matters because a tool only helps if people actually use it. The post doesn't detail how Foley Hoag rolled it out, so we won't guess. We'd just note that steady weekly use is the number a managing partner should watch.

Where billable hours fit in this picture

Any conversation about AI at a law firm runs into the billable hour, the unit firms use to charge clients for lawyers' time. If a week and a half of associate work takes a day, a firm has to think about how it prices that work.

We didn't see Harvey's post address pricing, and we won't invent an answer. Each firm will weigh that against its clients and its own model.

What we'd say is that associates aren't the problem here. Their judgment is still what clients pay for. The examples in Harvey's post describe AI handling the volume, while lawyers review the output and make the calls.

How we'd get a firm ready before buying

If we were sitting down with a managing partner, here's where we'd start. First, form a small group, lawyers plus someone from operations, to look at AI before any purchase. Munck Wilson Mandala had that kind of group in place early, per Harvey's post.

Second, write an AI use policy. Decide how you'll tell clients and how you'll get their permission, the way Munck's policy requires. Clients will ask, so the answer should exist before they do.

Third, pick one document-heavy job to test, such as a diligence review or a set of contract updates. Measure how long it takes today. Then run the test and measure again, with a lawyer checking every output.

Fourth, plan for adoption. Decide who trains people and how you'll track weekly use. Buying the tool is the easy part, and use is what shows whether it paid off.

If the test doesn't show a clear improvement, that's useful too. It tells you the job you picked may not suit AI yet, or that the setup needs work. Write down what you learned, adjust the policy or the training, and try a different job before deciding the tool isn't for your firm.

We've written before about taking a careful first step with AI on a personal injury file and about AI writing the first draft of a legal brief. Both make the same point: a person stays responsible for the work.

Law firm AI readiness FAQ

What is Harvey?

Harvey is a company that sells AI software to law firms for work like document review and drafting. On its blog, Harvey has shared examples from firms including Munck Wilson Mandala, Gray Reed, and Foley Hoag.

How to get a law firm ready for AI?

Start with a small group that evaluates AI, plus a written policy on telling clients and getting their permission. Then test one document-heavy job with a lawyer checking every output, and track how many people use the tool each week.

How much time can AI save a law firm?

According to Harvey's blog, Munck Wilson Mandala finished due diligence in a day that would have taken three associates a week and a half. That's one firm's account of one project, reported by the vendor, so treat it as an example rather than a promise.

If you want the long version of how we think about readiness, Start with a FREE quiz and Get Your AI Score.