Law Firms Are Buying AI: But Are They Really Using It?

On the eve of the annual ILTA conference this week, the legal consulting firm, Harbor, came out with a startling set of findings: law firms are buying AI but they don’t know what the hell to do with it.

And while there are some reasons to question the conclusions, the timing could not be better for conference attendees to ponder this yawning gap. Even though most of the attendees aren’t the ones empowered to change things. Like equity partners.

Some key statistics. The average spend on technology increased 41% from 2021 to 2025. Moreover, that spend is now 5% to 6% of overall firm revenue. Okay, that’s great, and suggests that law firms are finally embracing technology like never before.

But Not So Fast

But not so fast. Here is the critical finding: “No participating firm in Harbor’s 2026 legal lab reported having a mature framework for measuring AI’s business impact.” None. Zippo. 

As Harbor puts it:

Many firms are investing aggressively in AI without first defining the business and client outcomes those investments are intended to achieve. Without a strategic foundation, value cannot be measured consistently, investment decisions become reactive, effectiveness focuses on other tools rather than outcomes, and guardrails become more restrictive rather than enabling.

If true, that’s a pretty bleak picture. 

But There’s A Fly In The Harbor Ointment

Before we conclude it’s all doom and gloom for our dear law firms, there are some concerns about the findings. First, it’s not clear from where the information being reported comes. Harbor says it combined a “sounding board” of law firm CIOs, CTOs, and chief knowledge and innovation officers. However, the report does not reveal how many people were surveyed, or even if anyone was surveyed. 

We aren’t told which firms were involved, or their sizes. We don’t know if the 2021 and 2025 spending figures were from the same firms or whether the 41% figure is a base number or inflation adjusted. And the report could just as easily come from a roundtable discussion over bourbon rather than a statistically supported survey.

And of course, there’s always the definitional problem of what actually is software technology spending. Harbor doesn’t help us here either.

So, when reading the report, even though the numbers and conclusions sound valid, without obtaining the methodology and information about the data, we have to take them with a grain of salt. Just because you think something sounds true, doesn’t always mean it is.

AI Slop?

There’s another thing that bothers me: a fair amount of the report reads as if it was generated by AI. Take this sentence for example: “Technology alone is no longer a sustainable differentiator. Legal expertise alone is no longer sufficient. These are both table stakes.” There are other examples. 

The problem is not so much that an AI tool may have been used to help draft the report. It’s that without the methodology and underlying data, we can’t confirm whether the conclusions are AI generated or carefully thought-through conclusions based on data. That too raises a red flag.

Nevertheless, Harbor Raises Some Good Points. And Questions

Having said all that, there are some important points in the report that we should all think about. In fact, many of the things contained in the report are consistent with other findings and, along with systemic law firm limitations, suggest that a lot of law firms may be talking a good AI game but not much more. 

Harbor offers what it calls an Enterprising and Operating Model that identifies and tries to address these limitations. This model consists of four points discussed below. They all seem rather obvious and are standard innovation techniques, but the fact that law firms struggle with them tells you a lot. And to its credit, Harbor has resisted the temptation to do what many consultants do: take what’s simple and try to make it really complicated.

Measure What Matters

Harbor first correctly points out that without measuring value, there’s no way to determine whether an investment in AI is truly succeeding. As Harbor puts it, “Firms are measuring AI activity more effectively than AI impact.” Harbor suggests that firms need to determine how AI can contribute to profitability, efficiency, productivity, and outcomes. (All true). 

Invest With Intention

Harbor then suggests that after determining what success means, firms need to assess what investments are needed to achieve the desired outcomes. It also notes that, under the present decision-making process of most firms, the assessment of the technology ends once it’s purchased. It’s like the shiny new toy that everyone wanted but a few months later no one can remember why.

As Harbor puts it, the question is not just whether a project needs to be approved. The question is really whether it is the “right investment based on strategic value, risk, organizational capacity and long-term ownership.”  

Change Behavior Not Just Technology

Harbor opines that the issue firms need to address is, having invested in technology, how that technology can be used to fundamentally change how people do work. The key is not just more rote training with lawyers checking their emails while someone drones on about features and benefits; it’s being able to make lawyers and legal professionals see how AI can be used in their day-to-day work and make that work better. Otherwise, AI tools will just flop around like a hooked fish pulled up on dry land.

Build Trust At The Speed Of Innovation

Harbor points out that lots of firms have established AI acceptable-use policies. The problem is that many of these firms stop there. They fail to turn those policies into practical guardrails for their professionals so that they can confidently use the tools. Harbor also suggests guidelines need to be embedded directly in the way professionals do their work. Lastly, Harbor cautions that, since AI tools are accelerating, lawyers are experimenting with them somewhat on their own. And practice groups are developing their own AI workflows. That creates risk. 

But Harbor May Still Be Right 

Again, without the underlying data, it’s hard to wholesale embrace Harbor’s conclusions. Again, the report sounds right, but it doesn’t offer statistics on how many of the recommendations are already adopted and by how many firms. Nevertheless, there are reasons why Harbor’s conclusions may be spot on.

First, they are consistent with other findings. Several months ago, I attended Legal Geek in Chicago, where Gina Passarella offered many of the same conclusions. Law firms are buying and buying but haven’t figured out how to use AI to benefit their clients. Other reports are full of data that say roughly the same thing.

Second, while I hate to harp on it, the fact is the billable hour business model creates an underlying tension with trying to achieve success with AI. It’s hard for law firms to figure out how to measure value and success with tools that bump up against an ultimate reality: AI will take away billable hours. AI saves time. Lawyers sell time. It’s that simple.

Just last week, Thomson Reuters announced that its new AI platform will be able to perform at the level of a senior associate. Think about that: if it can do the work of a senior associate, then it could probably do the work of a junior associate and a paralegal. And that means the whole leverage model, which has made so many lawyers so much money, is threatened. Most firms simply haven’t figured out how to continue to be as profitable as they were before AI. 

Because the billable hour model is so ingrained in the culture of these law firms and there is no way to easily figure out how to recoup those lost hours without junking the model, leadership runs around in circles. Firms are left with purchasing AI and technology for purchasing’s sake, in hopes of impressing clients, without having figured out how to use it profitably.

Feeding The Hungry Sharks

Harbor also gets at another structural reason that law firms may be struggling with AI adoption. It is consensus-based decision-making. Harbor puts it this way: “For decades, firms optimized for delegation and consensus, quality, and risk mitigation … Decision cycles can no longer be measured in quarters or years, yet many governance processes have not changed to keep pace with the continuously changing environment.” And committee-based approval just gums up the works.

One other point, which in my experience here is absolutely correct, is that “many firms found themselves reacting to the loudest requests rather than intentionally investing in the initiatives most likely to advance firm strategy.” 

Lawyers are notoriously independent, and many partners are hard to restrain. When a partner with a $5 million book of business demands a certain tool, they are likely to get it, particularly since they can leave the firm at the drop of a hat. 

Moreover, law firms are made up of siloed practice groups, and the tools needed are looked at through that prism. So, technology and AI decisions are not made holistically for the good of the firm. They’re made to feed hungry sharks. 

At The End Of The Day, It’s The Clients

The final reason for the poor use of AI is the clients. Harbor’s report says clients are demanding better use and that law firms change and create real value with AI technology tools. 

But one thing I learned over almost 40 years of practicing law is that when clients do demand something, law firms can become remarkably innovative quickly. That firms are unable or unwilling to demonstrate valid measurements of value based on the AI tools they are buying can only mean that clients are not insisting upon it. Until they do, millionaire lawyers are not likely to change their ways.

Some Things To Ponder At ILTA

While there are some fundamental concerns with the Report and its lack of citations to data and statistics, many of the things it’s saying may very well be true. That’s not particularly surprising and makes some sense. 

So, while everyone at ILTA is in a closet talking to themselves about all the wonders of AI, perhaps they should read the Harbor Report and ask themselves: Is this the story of my law firm? If it is, then no amount of AI investment is going to move the needle much. 

Until clients vote with their feet. But then it will be too late.

---

Stephen Embry is a lawyer, speaker, blogger, and writer. He publishes TechLaw Crossroads, a blog devoted to the examination of the tension between technology, the law, and the practice of law.