AI Can Do My Job. I Quit.
It can probably do yours, too.
Like many New Yorkers, I’ve escaped the chill of New York to find some warm weather in Miami. And sitting on the beach, I had a revelation: Artificial Intelligence can do most of what I do as a fractional general counsel. So I quit. What I’ll do next is anyone’s guess, but in the meantime I’m pretty happy in my tent on the beach. A tent is the only option: I can’t afford a hotel because I won’t be able to work for a while, if ever, and there is no severance when you quit your own company.
I actually thought this through. Like many lawyers, I made my money drafting commercial contracts, leases, purchase and sale agreements, financing documents, private placement memoranda, compensation agreements, stock option plans, board resolutions, etc., etc., and that work is pretty much gone. It was nice while it lasted. And my litigation practice is on the way out too: AI can research, draft, and respond to motions and document demands in a fraction of the time it used to take. The only thing it can’t do is stand up in court and make an argument, or examine or cross-examine a witness. But I only do that a few times a year, and when I look at it, courtroom work and depositions account for less than 10% of my billables. And that’s coming soon, so I’m told. The robots, being owned or leased from some big tech company, will probably have better suits than I do. Yep, time to quit.
I used to think—like, last week—that I was OK because I’m not hired for drafting. That work is commodified anyhow and offloaded to lower-cost associates or offshore/low-cost legal service providers (big-ish firms call this “leverage,” and they charge a fortune for the work their associates and LSPs churn out—it’s the number one reason why my clients used to use me instead of a big-ish firm for much of this work). No, not me. I was hired for structuring and strategy. And AI can’t possibly do that, right? Wrong.
Turns out the reason AI can’t really structure or strategize well at the moment is actually by design: the confidence level to suggest litigation strategy and transaction structure is just not high enough for the AI system to confidently predict what to do. So it makes some suggestions, many of which are wildly wrong and often based on “hallucinated” information. But having worked with the tech for a while, I know that with some prompting you can (1) check any “fact” it comes up with and (2) get it to tell you what evaluation criteria you should use to make a decision, and then you can replug that evaluation criteria into the chatbot, which will allow it to make a confident suggestion. Eventually, AI will figure this out. That is the nature of the beast.
But I should have been fine, right, because 95% of what I do does not end with a written document—or a strategy or structure. That is just the beginning, actually. Almost always, there are other lawyers involved in the transaction or dispute, and/or human stakeholders at my clients who need to read and understand the document, structure, or strategy.
Because there are humans (still) involved, I tend to draft—often with the help of AI—long-ish (privileged, when it’s to a client) emails explaining the document, structure, or strategy, and then have follow-up calls. I suspect that, much of the time, the other human plugs my email into their own AI with a very simple prompt like “tell me what Jesse is saying” or “tell me what Jesse is saying and draft some questions and counterpoints.” Or, in the case of some of my more aggressive opposing counsel: “Jesse is an a-hole and a really crappy lawyer—rip his arguments to shreds and tell him how dumb he is.”
And here is the real reason I quit: it’s only a matter of time before the AI systems start to talk to each other. Why have I written an email or prepared a document only to have the human on the other side enter the email or document into their own AI system and then send something back to me responding? What’s the point of that? Why can’t two AI systems talk to each other, perhaps solve the dispute, or agree on all the transaction structure and draft all the documents—then send them to a human for an electronic signature? And then have the AI move money or file the necessary documents to resolve the dispute or settle the transaction?
There is a separate question of whether these types of systems can be owned by non-lawyers—which, if you think about it, is the only real question left once the tech gets to that level. As I’ve written before, ownership of those systems by lawyers or “Alternative Business Structures” is probably the most impactful discussion for the future of the profession, and it’s the one we’re going to be having for the next few years.
Putting aside the ownership issues, I have no doubt that the technology will get to the point where lots of legal work can be automated: two AI systems talking to each other, with minimal human intervention. I’m actually undecided on whether the lawyers managing these systems need to be high-cost senior specialists or low-cost junior lawyers. Either way, there will be fewer lawyers doing the work because so much will be automated.
After doing this for 20 years, I can confidently say there is nothing precious about legal work, or the vaunted “judgment” of human lawyers—which, in my experience, is often as wrong as it is right. The future is fewer, and more productive, lawyers. So I quit.
But wait. Maybe I should reconsider.
The question we, as humans, need to ask is: would that really be better than the current system? Do you really want two AI systems (either owned by a lawyer or some tech company) negotiating deals for you with minimal human supervision? Resolving disputes? Determining whether you’re in compliance with regulations—presumably determined by other AI systems? Or deciding whether you’re guilty or innocent (or whether you should be charged in the first place)? Maybe you do. If so, consider yourself a proud techno-utopian who is building a better future with the information you have today. Not a bad thing—but sometimes it misfires. The rest of us can happily live in the real world.
And the real world offers some real-world challenges. First, there are only so many large language models, and only so many companies that are going to make them legal-specific. Let’s say that, when the industry matures, there will be maybe 100 AI systems capable of handling high-quality legal work, and 10,000 “AI-native” law firms are really just wrappers on those systems. I don’t know if that’s the right number, but it’s unrealistic to think there will be anywhere near the number of legal systems as there are U.S. law firms (firms, not lawyers) today (463K): https://www.statista.com/statistics/822025/us-legal-services-market-law-firms/?srsltid=AfmBOoqaQ4xe_ecwuVjIa6Fuph8l-UhgQ-ng-DH4ussUp5RqG4FrYH5z. In any event, there will be far fewer lawyers. Like I said before, the future is fewer, more productive law firms.
When you have such a consolidated industry, there are serious issues with conflicts, costs, and quality. Explaining them here will sound persnickety, and, as someone who works with tech companies every day of the week, I can tell you that it’s all solvable. But the solutions are not obvious, and it’s very expensive to solve.
That’s where the AI vision of the future of law tends to fall apart. Once the VC/investor money runs out, AI lawyers are going to cost as much as the human lawyers they are trying to replace. It’s just that the money won’t flow to lawyers—it will flow to the people who own the AI (again, the debate here is about whether only lawyers can own law firms). People and companies with gobs of money to spend on lawyers—or the government, which has more money than anyone—already have top-shelf legal services. The current system works fine for them. Maybe there will be some productivity gains, but nothing that moves the needle. And for those high-end legal services customers (“clients”), I’m pretty sure the AI replacement is going to be inferior to what they have today.
But for the 99.9% of people and businesses that need to decide between spending money on legal services or something else, AI offers solutions—but an AI lawyer is not the answer. What they need is a great human lawyer deploying AI to deliver faster and cheaper legal services.
Your Fractional General Counsel is a law firm that uses AI to compete on price and speed without sacrificing quality. If that’s also your vision of the future, and you want to help build a law firm that uses AI to allow human lawyers to offer cheaper legal services, please get in touch with me. That’s what I’m interested in because the cost of legal services is too damn high. And, of course, the goal is to make up the lost billables by doing volume. So to help, you can also send me some business :). That last bit was legal advertising.
You may have noticed that this newsletter had a conjoined title. Yes, AI can do my job. But no, I won’t quit. Besides, I hate camping.
If your looking the future of legal services in the AI era, its already here: using AI to service my clients faster and cheaper. LFG.
Keep thinking, keep building,
Jesse
Hi, and welcome to my newsletter! I’m Jesse Strauss, Your Fractional General Counsel. I’m a lawyer with a private practice based in New York City, helping clients in the United States and globally with their U.S. legal needs. My expertise spans various areas, including raising funding rounds, employment issues, negotiating master service agreements, intellectual property, compliance, legal process management, and dispute resolution. My focus is on founding and nurturing great companies from seed to exit. Discover more at Your Fractional GC and book a complimentary 30-minute consultation. You can also follow me on Threads @lawyerjesse1977, on BlueSky @lawyerjesse.bsky.social, subscribe to my Substack here (follow me on notes), and follow me on LinkedIn here.