"Hey, bro, remember that referral fee thing I was telling you about? I drafted up a quick contract for it. What sections is it missing?"
“Hey, Dom! Please see attached for the template employment agreement. If this looks good to you then I’ll go ahead and drop the specifics for the employees.”
“Hey Dominick can you please look this services contract over and make adjustments as needed and then make it look pretty and professional?”
“One of my cofounders had some questions and feedback on the drafts you sent from his legal review. Will you take a look and respond?”
As a corporate and transactional lawyer who routinely interacts with entrepreneurs and scrappy start-up companies, these sorts of questions and requests are becoming increasingly routine. Armed with the inexpensive (or free) help of artificial intelligence (“AI”) tools like ChatGPT, Grok, and their competitors, the smaller-company community can DIY contracts and legal review at a speed and sophistication that was unimaginable just a couple of years ago. (Well, maybe not unimaginable. But, a couple of years ago, you’d expect a whole lot more typos and formatting issues in a non-lawyer-produced first draft of a contract.)
Of course, in virtually all of the cases I’ve encountered in my practice, what the AI tool produced was close, but not exactly the cigar sought by the user. The referral fee agreement involved two Arizona residents with no good reason for Wyoming law and venue to apply, plus it failed to address the effect of termination on the obligation to pay a referral fee. The employment agreement template did not address company ownership of employee-created intellectual property and contained an overly vague non-solicitation provision. The services contract lacked detail as to the scope of the services to be provided and had loose payment terms (although, I thought it was “pretty” enough). The legal review of one of the governance documents reviewed by the co-founder correctly raised issues that would be troubling, except that another document in the set assuaged the concerns raised completely.
At this stage of my career, I can identify these deficiencies and communicate them to a client. And, I think, when I do, the client sees the value-add of my independent assessment of the AI tool’s work product.
But just how value-add am I, or any of my transactional brethren? Sometimes, it’s hard to say.
As I have explained to some folks before, the AI-generated contracts and legal analyses are not necessarily bad; often times, the AI tool’s work product just lacks depth and detail or nuance and style that is honed over time through experience. The legal ability of the AI tool, in terms of adequacy and accuracy, is more closely akin to the work product you might receive from a new lawyer fresh out of law school than a lawyer with several years of relevant training and experience. AI tools don't yet possess meaningful wisdom.
As a lawyer who is closing in on a decade of full-time practice, do my comments and revisions to AI-generated contracts materially enhance my client’s position and protect their interests? In many cases,
I think the answer is “yes.” I’m fairly certain that if there’s a dispute over the referral fee contract mentioned above that my friend will be happy not to have to litigate in Wyoming. And I think the value I add in these situations exceeds my cost to a client, more often than not.
At least, for now.
The reality is that usually the best outcome for any client is for the quality of their contract to never be tested. A poorly-written and loose contract works just fine when the parties never end up in a dispute
over its terms; absent a dispute, the hindsight value of changing Wyoming venue to Arizona is zero at best. So, if you’re modeling out expected value-add, you have to place a value on the probability that
the improved language will ever be tested and multiply that figure by the value of the improved language, and then compare that product to the cost of the legal services. This is hard to do in practice, but theoretically it’s the only way to economically justify hiring a transactional lawyer at a particular cost.
As any lawyer with a year or more of experience will tell you, practice doesn’t make perfect, but it certainly makes better. There is a reason you pay a law firm more for an experienced partner’s time
than an inexperienced associate’s time. But even though the junior associates are cheaper, they’re still mostly a heck of a lot more expensive than even the premium versions of the leading AI products
available to the masses. And the AI tools, like the junior lawyers, keep getting better with experience and training. In economic terms, the marginal comparative expected value-add of the human lawyer is
diminishing and will eventually become de minimis or worse.
This value curve trends down because, eventually, the AI tools will simply be better transactional lawyers than the best of us, and there’s no reason to believe they’ll be overly expensive like many
lawyers are. The AI tools can crunch more data than you or me, don’t need sleep, and don’t have a family or distractions from the tasks at hand. Someday (maybe a few years off, but eventually) the idea of employing a human to draft a contract for your small business deal will feel a lot like riding in a horse-drawn carriage, washing clothes in a bucket of soap, or renting a DVD: people will only do it if they value nostalgia over efficiency.
For transactional lawyers not enticed by the financial prospects of anachronism, the challenge will be to remain a value-add as AI tools catch up to their abilities. (Transactional lawyers paying attention
perhaps ought to feel a lot like 15th century scribes and manuscript copiers learning of the catching-on of the printing press.)
Much of what has led to the economic success of transactional lawyers as a profession has, in my view, been driven historically by the increasingly complex and technical nature of contracts and
legalese, where the unwary and untrained are wont to hire someone with the knowledge and understanding of these complex documents and sentences to protect them. My view is that the AI tools are in the process of diminishing whatever value transactional lawyers have historically enjoyed with respect to the technical side of the practice.
I don’t grieve the erosion of transaction costs even at the cost of my own skillset, however, because I think transactional lawyers and value-add will instead have to take the form of the lawyer serving
as best he or she can in the lawyer’s most basic and non-technical role: as a wise counselor.
To be a truly effective counselor—to help guide someone through a decision-making process, to weigh pros and cons, to help identify and understand the all-things-considered best approach to choosing
among options, to appreciate risk and reward, to ultimately help clients achieve a sense of confidence that they are making an informed, mature decision—requires empathy, trust, and genuine human
connection. Less technical, more philosophical. Less talk about what the contract says and more about how it makes the client feel. It will take longer for the transactional lawyer as counselor to be displaced
by AI tools. Perhaps not forever, but certainly longer.
The transactional lawyers who can use AI tools to do the technical work and use people skills to counsel clients will have enough time to keep a career going and earn a living adding real value for a while longer.
And when the AI tools get good at counseling clients, like real people? Well, few mourn the demise of the scribes and manuscript copiers.
Click here to view the Fall 2025 article published by The Federal Lawyer.
Published with permission from The Federal Lawyer, a magazine published by the Federal Bar Association.
about the author
Dom San Angelo practices business and transactional law, serves on the Board of Directors of RideNow Group, Inc., and is an adjunct professor of business at Grand Canyon University in Phoenix, Arizona. He graduated with a Bachelor of Arts and a Bachelor of Science in Business Administration from The University of Arizona, as well as a Master of Arts in Philosophy from Duke University and a Juris Doctor from Duke Law.