The decline of time-based law firms

Lawyers' billed hours constitute the inventory of law firms and the foundation of their profits. Legally trained AI will decimate those hours. What happens then?

Feeling apocalyptic about the impact of legally trained Large Language Models (LLMs) on the future of law firms? No? Allow this excerpt from Corporate Counsel to help you with that.

[T]here’s reason to believe a boatload of tasks could be replaced by AI—about 44% of legal tasks within the U.S., according to Goldman Sachs’ report on generative AI. If that comes to pass, “the next thing that happens is GCs saying we’re not paying $900 an hour for associates to do something a machine can do for 2% of the cost,” Cemenska said. “If 44% of the work goes away, about 85-90% of junior associate work goes away and a significant portion of senior associate and non-equity partner work is going to be partially eroded.”

Now, to be clear, that calamitous estimate is based on a Goldman Sachs “report” that runs all of 20 pages and purports to cover vast swathes of global industry. The only lawyer activities mentioned in that report are: “Review documents and proposed actions for compliance with legal, regulatory, and corporate standards; provide arguments and scenarios for and against compliance in unclear cases.” So I wouldn’t call it the definitive analysis of LLMs in the legal sector.

But the legal tasks enumerated in that report are also only a sample. Equally susceptible to LLM-ification are legal research, legislative and regulatory review, caselaw summary, memo writing, factum drafting, document creation, due diligence, and contract analysis, to name just the most common lawyer activities. ChatGPT4 alone could do a good first pass on most of these, never mind more advanced programs like Co-Counsel, Harvey, Lexis +AI, Jurisage, and many more to come.

What percentage of your law firm’s billed hours involve these and similar tasks? You should probably find out, because that’s the percentage of your firm’s current revenue stream that’s at risk. If even half of Goldman Sachs’s 44% estimate proved accurate, that would have a disastrous impact on law firms — effectively losing one day of billing out of every five.

Generative AI is an incredible time- and labour-saving invention for knowledge work. Law firms are businesses whose inventory is the time and labour of their knowledge workers. Generative AI cannot help but decimate law firms’ inventory. The new legal business tool and the old legal business model are fundamentally incompatible.

I don’t buy it, you might say. People have been prophesying the arrival of law firm killers for decades. First it was multi-disciplinary practices, then the Big 4 accounting firms, then document automation and predictive coding, and then alternative legal services providers. Every one of these challengers came at the king and missed. Why would this time be any different?

The difference this time is that, unlike other legal technologies and innovations that came before it, generative AI is not an expert system. There’s no professional or technical qualification required to use it. There’s no costly infrastructure or regulatory barrier to entry. There’s hardly any admission fee. People with little or no legal background can ask a legally trained LLM questions or request documents using everyday language and get what they need in seconds. They don’t need a lawyer to act as an interlocutor between themselves and their legal solution.

The biggest threat generative AI poses to law firms is that it will disintermediate lawyers from many legal solutions. In doing so, it will empower end users and invert demand for legal services. People and businesses that would normally turn to lawyers for legal services will use LLMs instead, either for the initial steps on their legal journey or for the whole damn trip. AI won’t take away all the hours. But it’ll take away enough to do some serious damage.

This applies especially to corporate clients. An in-house lawyer at a financial institution once told me about the 500 unread emails in her inbox that morning, and how none of them were going to get opened that day because she had too much other urgent work to do. That’s endemic throughout the law department world.

I can pretty much guarantee that thousands of corporate counsel are using ChatGPT4 right now to triage their most basic work. As legally trained LLMs become more affordable and reliable, they’ll become an indispensable in-house tool. That’s why I think the question of whether law firms will adopt legal AI might not be as critical as it seems. Eventually, most firms will integrate legal LLMs into their workflow — but if clients can access those same tools, then they might end up saying to law firms roughly what Chili Palmer said to Bo Catlett.

Now, this will be fantastic for the millions of people who’ll use LLMs to answer basic legal questions and even discover previously unrecognized legal needs just by asking an LLM what they should do about their problem. As I wrote last week, I think individual lawyers could benefit, too, from a much larger pool of clients whose basic needs have been met and who have developed more advanced, human-centred legal needs that require lawyers’ judgment, empathy, engagement, and strategy. I can see a win-win there.

But law firms? Lawyer hours are essential to their business model. An individual lawyer doesn’t need to sell their time in order to make money. A law firm doesn’t know any other way.

Firms generate profits by selling the time and effort of their lawyers for more than those lawyers are paid. The more hours a lawyer works, the more profit the owner makes. The fewer hours, the less profit. Generative AI is going to drive a mass re-direction of legal demand away from “hours of lawyer effort” and towards “seconds of machine activity.” That’s not exactly optimal for the firms.

Jacking up rates can only go so far to make up a loss on this scale — and in any event, clients are already learning to push back on aggressive rate increases and to look for alternatives in smaller firms. But the challenge goes beyond that.

Law firms price their work not according to the job being performed or the client value produced, but based on the lawyer who’s performing it. A lawyer who spends an hour on a mundane task and spends the next hour coming up with a brilliant strategy charges both hours at the same rate, and the second hour is a huge bargain for the client. (The system penalizes lawyers as they become more skilled, because their annual rate increases lag behind the ever-higher-impact value they provide.)

If law firms sold their services according to the value those services delivered, or even on a simple fixed-fee basis across the board, the arrival of LLMs would require little adjustment and would mostly enhance profitability. But they don’t. They sell hours of effort, and LLMs diminish those hours, and law firms have a problem.

How do they solve this problem? It’s facile to say firms should just switch to a value-pricing or market-pricing business model — you might as well suggest changing the engine on a semi-trailer while it’s barreling down the highway. “Hours worked” is baked into the DNA of most law firms: In addition to determining how services are priced and billed, it’s also how lawyers are compensated, evaluated, and promoted. The only, and I mean the only, metric that every law firm in the world tracks is “hours billed to clients.” The traditional law firm depends on lawyer hours.

But lawyer hours are going into steep decline. And law firms have only ever had one response to declines in demand for lawyers’ hours: to lay off lawyers. We’re in one such wave right now.

The reduction in law firms’ lawyer workforce that will follow the proliferation of legally trained LLMs, though, could go beyond anything we’ve seen before. Firms in the aftermath would be smaller, but could eventually return to previous levels of profitability by cranking up rates and working the remaining partners even harder.

Is this my advice to law firms? No. My advice is the same as it’s always been: Deepen your relationships with your most desired clients so that you can understand and anticipate their multi-dimensional needs and collaboratively develop systems and solutions to address them, charging a fixed annual fee with variable pricing options for unexpected, unique, or particularly complex matters. Launch that new system in a separate business entity and then gradually migrate your work and clients from the old system to the new one. There. Saved you a $60K consulting engagement.

If you’re in a law firm today, ask yourself which of these approaches seems more sensible, responsible, and visionary to you. Ask your colleagues the same thing. And then decide what you’re going to do about it.

You have, ironically, time to figure this stuff out. Legally trained LLMs are still in their infancy, and corporate law departments’ current hesitancy to adopt this technology buys you more breathing space again. This movie isn’t half-started yet. But the opening credits have rolled and the plot is taking shape. Your firm is about to enter the action. What’s your first line of dialogue going to be?