Mike Spivey: Welcome to Status Check with Spivey, where we talk about life, law school, law school admissions, a little bit of everything. I’m Mike Spivey; I’m the CEO and Founder of the Spivey Consulting Group. Today, I’m with someone I’ve known for 10 years. He’s a former world record holder; I’ll keep the suspense in there, and you can look it up in our show notes. Vincent Sheu.
Vincent went to UC Berkeley and received multiple degrees, statistics and molecular and cell biology. He then went on and got a master’s in bioengineering. He took that to Stanford University, not just for a JD, but also an MS in computer science. What does someone do with all of that? Vincent not only interned as a 1L at Perkins Coie and as a 3L for Davis Polk in New York City, he spent a summer as a 2L with the NSA in their general counsel’s office. So he’s seen a good bit.
And he’s seen a good bit on the cutting-edge side, which is why Vincent was a first mover in not only using AI in his legal work—he was a general counsel, and he basically had AI being his entire GC team—he codes AI. So Vincent has probably been one of the first movers on the planet in, how do you use AI to make what you do as a lawyer better for your company or your clients, while not infringing, impeding on what you and the humans at your company and client do?
Vincent’s a great guy, has a high EQ, and he cares a lot about people, and I think he’s going to use AI the right way to help people do things in their lives without replacing the connectivity in their lives. There’s a lot connectivity on this podcast. Without further delay, this is me and Vincent.
Vincent, it’s been a while. We caught up this week. It’s great to catch up.
Vincent Sheu: Yeah, great to see you again.
[2:05] Mike: Yeah. What are you doing? You have like a million hobbies, world records, and passions.
Vincent: No longer a world record holder. For context, what Mike’s referring to is, I used to be a competitive Rubik’s Cube solver. And so back in the day, I taught a class at Cal to 150 students a semester, I organized a couple world championships and national championships, and also set a couple of world records. But now I am retired. Although, I still do, actually, legal work for the national competitive body for Rubik’s cubing.
Mike: So there’s two things. One, are you still number one on Reddit’s funny for the—
Vincent: No. No, no, no, no, no. I was a caption contest for Above the Law at one point, though.
Mike: Was that the clip of Bill Walton talking at the Cal/Stanford game and you—
Vincent: It was, yeah.
Mike: You’re behind him, right?
Vincent: Funny story, that actually was a miscut. That scene he was talking over was prerecorded before the game. That’s why we were studying. Because this is back when I was in law school, so at Stanford, and we went to the game because I had known Bill from my days at Cal. And we had a chat, and then I was studying by the court because it was, like, an hour before the game. They then spliced that in during the game to make it look like we were actually studying during the game, which is funny.
Mike: Is it still—I couldn’t find it. We’ll link it if it’s still up.
Vincent: Yeah, I can go find it. It’s somewhere out there.
Mike: It made SportsCenter.
Vincent: Yes, yeah.
Mike: All right, so you’ve been on SportsCenter. Did you solve the Rubik’s Cube in 0.96 seconds? Is that even doable?
Vincent: That’s the two by two. That was one of the world records I held.
Mike: What’s the fastest you ever did a Rubik’s Cube in?
Vincent: 8.55, which back then was considered quite fast, and now it is not quite as fast as the best in the world, at this point.
Mike: I think our listeners tuned in to learn about the Rubik’s Cube, but if you want to shift to law school admissions…
Vincent: Law school admissions is more fun.
Mike: You applied more than 10 years ago. Is that correct?
Vincent: Yeah, I applied in 2013 as a senior coming out of Cal at the time. I didn’t know what I wanted to do, and so I actually applied to a combination of graduate programs and law school programs. And also my stats were kind of on the odd side, and you know, I had a, shall we say, a lower than average GPA, considering I was coming from a STEM field. And so I applied to about 25 schools.
Mike: Okay. So your major was pretty dynamic, and I know you shifted through a lot of different major possibilities. What was your precise major?
Vincent: So I ended up doubling in statistics and molecular and cell biology out of Cal.
Mike: And I’m going to guess your LSAT was above most schools’ medians or all schools’ medians.
Vincent: Yeah, it was a 99th percentile score.
Mike: What made you pick those 25—you don’t have to go through all 25.
Vincent: Oh, yeah, yeah.
Mike: Seven to nine is the typical range. Although even that’s a little misleading, because there’s a lot of people that apply to just one or two, and then there’s, like, a gap, and a lot of people that apply to five to seven, a gap. 25 is a lot. But what was your strategy? What were your anxieties, concerns? How did you end up at Stanford?
[4:30] Vincent: Yeah. The anxiety I mostly had was, one, I didn’t actually know what I wanted to do exactly in the law. Like I had some inklings towards science policy; that was what I was focused on at the time. But also, given my numbers—and I’m going to use some terminology that we used back then. I don’t know if it’s still current, so Mike might have to correct me here. But I was an extreme splitter, so I was far below the 25th for most of the top 14 schools, for example, and even some of the top 20 schools, while my LSAT was above median or above 75th for a lot of the schools. And so I really had no idea where I was going to end up or how things were going to land, and so that kind of necessitated applying to a much wider net than most people would.
Mike: You were an extreme splitter—you were a super splitter.
Vincent: Yeah, super splitter.
Mike: So high LSAT, lower GPA, 25 schools. Was Stanford your top choice all along?
Vincent: Stanford was by far my top choice, and that was because I was technically minded. I was also focused on law school, and so there wasn’t a better school out there anywhere in the world that had that combination.
Mike: Well, let’s focus on Stanford. Did you visit the school?
Vincent: Yes. Yeah, so this is actually before I even received the admission decision, but for context, I’m actually local to the Bay Area, and so that’s where I grew up. My parents actually brought me in a basket to Stanford to visit their friend who was doing a PhD in electrical engineering at the time. So this is back in the ‘90s, if that tells you how old I am. Literally, I was like less than a year old; they carried me in a basket.
Mike: I can’t believe that’s not on your Wikipedia page. Go on.
Vincent: I actually don’t even think the Stanford Law School admissions people even know that, so I’ll have to tell Faye that sometime.
But yeah, so it was kind of the default place that we would also take guests and people visiting out of town, too, just to walk around. So I was already pretty familiar with the school, but at the time, I’d never actually wandered over to the law school side. So I emailed the admissions office at the time and just asked to set up a quick visit. I was curious about, you know, sitting in on classes, meeting some people, just getting a sense of what school life was actually like. And so I did that kind of in the spring. I picked kind of a quieter time when they weren’t as busy with admissions, because the admissions office is overwhelmed, at points, with just the volume of applications this year. But they brought me in. I had a chance to chat with a bunch of people, professors, students, and admissions office people, and it was a great experience.
Mike: So, you and I didn’t know each other then, correct?
Vincent: Not—yeah, I think we had just met at the time.
Mike: Right. You did the exact right thing. You visited when you could get people’s attention. I know you stayed in touch with Faye. Did you get to meet Dean Deal?
Vincent: That time, yes. I wouldn’t expect to. She’s an extraordinarily busy woman, but I lucked out at that time, and I just happened to say hi to her, and we’ve actually remained friends ever since.
Mike: And that’s rare. It happens, if you catch lightning in a bottle. I wouldn’t strategically plan—
Vincent: No, no, no. Do not do this strategically to become—like, I like her as a person, and that’s why we’re friends.
Mike: Right. Do you remember your personal statement topic?
Vincent: Vaguely. So I actually repurposed part of my college personal statement. And I basically continued that on. I think I might have said something along the lines of like, this is what I wanted to go into college. This is what I’ve kind of gained from that, and now this is the new direction that flows from that. And so I was trying to create a pretty clear throughline between all the different things in my life and why that actually led to law school being a very logical next step for me.
Mike: So, as a super splitter, you have a personal statement that’s differentiated, you have a background that’s differentiated. If you remember, how many schools admitted you? How many schools waitlist-admitted you, how many schools waitlist-denied you, how many schools denied you? And you can just do admit/deny if it’s easier.
[7:33] Vincent: I received admission decisions at 23 out of 25 schools.
Mike: Okay. As a super slitter.
Vincent: As a super splitter.
Mike: So let me tell you what you had going for you. You had Berkeley, great undergrad school with great brand recognition. When you say you went to Cal, how many people don’t know that that’s Berkeley?
Vincent: Oh, it’s a major issue for the school, like the brand. Because Cal is the sports and what we compete under, but it’s the original University of California, and that’s why they compete under Cal. But yes, Berkeley would be the term on the academic side.
Mike: Having done this 26 years, having been in charge of marketing and branding for two of three law schools, I had a feeling that’s a major issue for the school.
Vincent: Mm-hmm.
Mike: Right.
Vincent: Yeah, that’s okay, though.
Mike: Yeah it’s okay. So, you went to Cal Berkeley. You had a highly differentiated two majors on your resume. You had a highly differentiated resume; I’ve seen it. Not just in undergrad but in law school, you were deeply involved in a lot of organizations, right?
Vincent: Yeah. I’m an overcommiter. And whenever people ask, “Hey, can you do something for us?” I typically say yes. I just like to be helpful in that respect. And so that ended up with me just essentially taking leadership positions and doing a lot of work for a bunch of different organizations on campus.
Mike: So admissions always boils down to one word for me. Differentiation. You differentiated with a lot of things. A 180 or a 179, 178 LSAT. I mean, by definition, that differentiates. A bunch of leadership in organizations. Majors that differentiate, a school that differentiates. I’m not surprised, in the era that you applied to, you went 23 of 25 at all. I have feeling your personal statement was exceptional.
You might be surprised to know that, if you applied this year, LSAT scores have been greatly diminished not just in U.S. News rankings, but in the downstream consequences, which are LSAT scores are now half of the weight of which they were, but also employment is more than double the weight of which it was. So interviews, which I think you would do well at, are taking much more precedence and admissions. LSAT scores, over the last two to three years, you can see admissions offices—I used to not use this word because it was overused, but they’re being more holistic. And more holistic means they’re not just infatuated with this 179 or whatever LSAT. Does that make sense?
Vincent: Yeah, absolutely. Especially because the recruiting timelines for law firms have been moved so much. Like at this point, when you recruit, you don’t actually have much more than maybe your fall grades.
Mike: Not even fall grades. We added a whole department, someone who did hiring for biglaw and midlaw who’s leading up a new department at our firm to help people with that chaotic recruitment process. Because law firms are reaching out to students, if not the day they start law school, before they even start law school. It is crazy.
Vincent: Yeah. So they’ll give you conditional offers, and basically you have to maintain, you know, this sort of like academic preparedness. But at that point, the only data points they might have might just be your law school application.
Mike: Law school application, undergraduate GPA maybe.
Vincent: Maybe the LSAT score.
Mike: They’re looking at LSAT score. It’s a whole new game. This is not in my arena. I was a dean of career services, but we’re adapting to this ever-changing world, too.
How long did it take for Stanford to admit you?
Vincent: I was admitted, I believe, on April 6th. Either the 6th or the 8th.
Mike: When did you apply?
Vincent: I applied, I think, in November. I would say I was one of the later regular decision decisions. It was pretty far into the process.
Mike: So you had a clear number one. They’re holding on, holding on. How was the waiting on them for you? Because I’ve seen it cut a lot of ways, but one of the ways is pure anxiety.
Vincent: Yeah, I think, this is one of those things where—if you remember undergraduate admissions, they all come out pretty much at the same time. Law school is different because everybody does rolling at this point, right? But if you look at the timeline for undergraduate admissions, it was basically at the same timeline.
Mike: It wasn’t rough on you.
Vincent: Yeah. I think at that point, I’d set my expectations sufficiently… we won’t say low, but, like, I knew my position, right? I knew that it wasn’t likely for me to be getting every single one of my top choices, given my GPA at the time. And so, you know, it was a nice surprise and something that I really welcomed when I got that call.
[11:10] Mike: How did they do it? Was it a phone call?
Vincent: It was a phone call. I remember exactly where I was, actually. I was at the Presidio near the Golden Gate Bridge.
Mike: We remember the things that are emotionally the most important to us.
Vincent: Yeah.
Mike: Not surprised at all. You probably remember the conversation, even.
Vincent: Yeah, I was giving directions to my friend, and then, like, literally I saw the phone number come in, and I was like, sorry, you need to just pull over because I need to take this call.
Mike: Yeah. That’s a great story. You didn’t just get a JD right? You got a master’s.
Vincent: Mm-hmm. And computer science as well.
Mike: Is that part of that “can’t say no”? Or do you just have an active brain always seeking new things?
Vincent: So, back when I applied to law school, I had also applied to master’s and PhD programs, actually. And so, for example, for a couple of schools, I was considering joint offers with other fields ranging from biostats to biology to public health to even East Asian studies and a bunch of other things.
And so what I ended up doing was, I actually got the admit to the Berkeley bioengineering program as a master’s, and so I deferred Stanford first to do that. Then when I got to Stanford, I also did the joint degree with computer science.
Mike: Did you get the admit, and then ask for the deferral? Please tell me yes.
Vincent: Yes.
Mike: Okay.
Vincent: And it was a pretty tough decision, too, because I was like, it’s already there. But, you know, at the time in the direction I wanted to go in, specifically science policy, having a graduate education and training would’ve been helpful for the field I was hoping for.
Mike: Right. If you’re admitted and ask for a deferral—I mean, your reason is as good as it gets, that you’re already on campus. They’re not worried about you. But there’s many reasons, even not being on campus, that are good reasons. The key is to get the admit first, because then you have all the leverage. I have seen it go so south, which is why I asked you that question, when people say, “I’m applying to your law school, and I’m going to be asking for a deferral.”
Vincent: Yeah, because then they ask, why don’t you just apply next year?
Mike: Exactly. That they have no incentive to admit you. They would much rather see if you’re going to follow up in your commitment to them.
Vincent: Yeah. I was a bit naive, because I hadn’t studied law school admissions so closely at the time, and so this came to me because I was basically like, “Well, I wonder if I could do that.” Like, I had found out about the admission on the bioengineering side, and so I was thinking, like, well, I’ll just try, and if they don’t let me, then I’ll just go to law school now. And it just so happens they let me.
And later I found out that, at least for many schools—I won’t speak for any one school in particular. But, like, the standard really is, like, if we’re already admitting you, and you’re going to go off and do something interesting in the year that we’re deferring you for, you’re going to come back an even stronger member of the class.
Mike: Yeah, and you did exactly that. And your master’s correlates to how you practice law. You had a plan going into law school. You kind of had an idea of what you wanted to do. Your master’s correlated with that. What was your plan?
Vincent: Yeah, at the time, so I come in with the molecular/cell biology and statistics undergrad. I knew I wanted to bridge that to a more engineering focus, and so bioengineering was, like, the next logical step that eventually culminated in the computer science side at Stanford as well.
At the time, I really wanted to do science policy. I think I had a recommender mention something along the lines of, “Oh, I hope Vincent becomes the head of the FDA at some point.” That was before I found out that usually the heads of the FDAs have MDs, and so that didn’t end up working out, but I wanted to work in the government. You know, the OSTP would’ve been a great place for me to spend a summer or two. It didn’t quite work out that way because of the way the kind of wind shifted with regards to the country, and so I ended up going into more startup tech instead. But that was the intent at the time.
[14:03] Mike: Tell us about that. What was your recruiting process like out of Stanford? Did you work at Fenwick?
Vincent: I did. Yeah. So I actually spent four years at Stanford, because I was doing the joint degree. And so I had three summers. And so for that first summer, at the time, that was the one where people just kind of said, you know, do whatever you want. I won’t quite say a throwaway summer, but it’s more like, you know, a place to kind of get your feet wet and get some experience.
And so I did my first summer as a 1L summer associate at Perkins Coie. I was primarily doing IP litigation, and that was probably because I had the science background coming in. It was a logical fit given my background. This is out in the Bay Area, and so there’s a lot of IP work, and so that’s what I was learning my first summer.
When I was recruiting for the second summer, that’s where I had some choices. So I recruited at a bunch of different places across the country, just kind of seeing what was out there to see what I could get. And because I had that third summer in mind, I knew I had another free summer. And so I recruited for firms, picked up an offer from a New York firm actually, and then deferred that, because I also picked up another offer from the federal government in my second summer. Knowing that I had that third summer—and usually that last summer before you graduate is the one that becomes your postgraduate offer—the firm said, okay, come back the next year, and we’ll take you as a summer associate then as well.
Mike: And where did you end up after Stanford?
Vincent: Yeah, so as a 2L summer, I went to the NSA through their general counsel’s office. That was a legal internship. Then, for the third summer, I split New York and Menlo Park for Davis Polk. Ultimately, I did receive a return offer, but ultimately decided not to go there for other personal reasons. And then I ended up at Fenwick as a first-year associate.
Mike: Okay, that’s what I thought. Clicking on a couple things, you worked for the NSA for summer.
Vincent: Yes.
Mike: So that’s in Maryland, it’s the National Security Agency. Did you have to sign away your First Amendment rights?
Vincent: I treat it like any other job, other than the fact that you are working for the federal government and there are things that you can’t disclose, but otherwise it’s just like any other job.
Mike: Okay. So, I have a family member who worked for much of his career at the NSA, so he has some interesting stories. He had to check his cell phone in at the gate, and you probably did, too.
Vincent: Yeah.
Mike: In and out—right. You mentioned IP litigation. One of my favorite meetings ever was, I was trying to get the CIA to recruit when I was Dean of Career Services. I haven’t just done admissions. I’ve done career services. And the CIA hires a lot of lawyers; that won’t surprise you at all.
Vincent: Mm-hmm.
Mike: And the person I met with was an IP lawyer at the CIA. I couldn’t meet at Langley for obvious reasons, so we met at the bar at the Mandarin Oriental Hotel. She said her name was Carrie Lionheart; what do you think the odds are that that was her name?
Vincent: Probably actually her real name, yeah.
Mike: Yeah, it could be. Maybe.
Vincent: You’re right, all of these organizations have large legal needs, obviously, given the work they do. And so they all have these legal honors programs and summer honors programs like the one I did. And the reason for that is because they need lawyers, essentially, to make sure that everything they’re doing is legal. Those lawyers have an external-facing role oftentimes, because, you know, they’re stepping into court, and so they really do need to use their real names.
Mike: Even in her arena—I mean, it wasn’t even litigation, it was just intellectual property law—they had a big need at the CIA, because think about the cutting-edge technology they’re developing. Everyone wants a piece of that or wants to know what they’re doing. That was a fascinating meeting I’ll never—just like your phone call, I’ll never forget meeting with the CIA at the Mandarin Oriental in DC.
[16:59] You moved on. You went to Fenwick. I want to shift gears to what I think you’re going to be an expert in, and what law schools are rapidly trying to get ahold of and catch up on. Obviously, you weren’t using what’s known as “AI” now, but I’m going to guess you were way ahead of the curve at using technology that essentially was artificial intelligence. You’re the expert; I’m not. Is that right, and if you fill in those gaps that I’m missing?
Vincent: Yeah, so at the time, what we considered “AI” was a little bit different from what we have now. I think, at the time, GPT-2 was barely out. It wasn’t really something that was in the mainstream, right? So the things that I was studying on the computer science side, like natural language processing, those techniques are actually, in many ways, informative of what we use today, but in many ways also a little bit out of date compared to what modern technology is.
So like, on the CS side, like, I was coding up models, and I was training models and things like that. And so, when I later brought that experience out to practice, Fenwick was a little bit ahead of the curve in terms of adopting technology, but I think, by and large, the attitude of your average biglaw firm was pretty much like, you know, “This is the way we’ve done things and this is the way that we look to do things in the future if we can.” Since then, I think most law firms have greatly shifted in terms of innovation, both on the technology side and on the business model side. And so, I would say that that was at the very beginning of what that was in terms of what’s possible.
I think on the Fenwick side, you know, they already had a lot of internal tooling that allowed people to cut down on a lot of, for example, document generation. That was something that was already being done internally, I think. On the side of review, like, that was probably a little bit more of a manual process at the time. It’s definitely not anymore. You kind of trusted technology to do small, discrete steps that were defined, that you knew they could repeat over and over again, as opposed to being more generative.
Mike: I’m going to guess the West Coast firms, particularly where you were, are a little bit ahead of the East Coast firms as far as embracing innovation. Is that accurate?
Vincent: That’s my impression.
Mike: Actually, how would you know?
Vincent: Yeah. Yeah, so I later on became a general counsel, right? That involved talking to a lot of external firms and seeing what their technology stack looked like. I would say that the West Coast firms oftentimes will be exposed to technology just a little bit earlier, because, you know, your Fenwicks and your Wilsons and your Cooleys and Gundersons, for example, are all working with the startups that then become the innovators that then sell products back to those law firms, right? And so they just have a little bit of exposure that’s earlier.
Mike: That makes sense. I think “exposure” is a better word than “embracing.”
[19:10] You mentioned you became a GC. That’s when you really said, “Hey, I’m going to start using this to help the ROI of our entire company.” Is that accurate? “I’m going to start using this tool of AI.”
Vincent: Yeah, at some point, you run into so much work that you just need that double-check, right? You don’t have time to sit down for another three hours just to review all of your work to make sure you didn’t get anything wrong, and so, adopting AI is basically almost necessary for you to function well in a professional environment, especially when people from every other department are already using AI.
Mike: What were your hours at biglaw versus your hours as a general counsel?
Vincent: It varies a lot. As we all know, biglaw charges by the hour for the most part, right? You’re billed by the hour, and so you do tend to work more hours than your average, let’s say, engineering job. General counsel roles really scale up and down just depending on the work you have. And so, there are definitely times where I was working much, much, much harder than biglaw, and there were other times where, you know, it was a little bit lighter than biglaw. It just kind of depended on what the needs were.
Mike: Okay. That’s relatable. As a CEO of a company—and you met Anna Hicks-Jaco; she’s the president of our admissions side. She has busier days than me; she does a lot on a daily basis that I can’t do. But I have days where I’m working 18 hours, 19 hours. That’s your GC day that blows up, right?
Vincent: Yeah. There’s nobody else to pass the buck to, right? Like, you’re not hiring an external firm, probably, because it’s not big enough, and you’re the top law person or might be the only law person. And so you better get the work done.
Mike: The way I once heard a basketball coach describe it was, “When I became a head coach versus an assistant, I had no one to look over my shoulder and ask, ‘Can we do this or not do this?’ anymore.”
I think our listeners are actually going to want you to guesstimate hour load, average day, how many days a week at biglaw versus GC. Because people are trying to, you know, make these decisions.
Vincent: Yeah. I was GC for smaller companies, right? Like I wasn’t GC for Google, obviously. And so the typical in-house job you’re going for is typically, you know, commercial counsel at a large company or something like that. Or maybe IP counsel or product counsel or something like that. Those jobs have more defined hours, although there are certain companies that say they work just as much as biglaw as well.
For me, there was nobody else to go to, as I mentioned. Right? And so, like, I didn’t have a senior counsel to report to, or a general counsel to report to. It was just me. And so I basically would take calls whenever. It would be very much like biglaw where, you know, I’d be out on a weekend, and the call would come in, and I’d have to step aside, take it, and be like, “You know, I need to work right now. Like, I can’t go out for this.”
Mike: Yeah. I’ll give you some hours from my end and from Anna’s end if it’s helpful for the listeners. I never take a day off. I don’t like getting behind. I always look at email. My lowest days would be an hour; my heftiest days would be 18 hours. If I had to guess, Anna’s typical day is 9 to 10 hours. And again, we have days that are less, and then we have some blow-up days that are all hands on deck. Is that relatable?
Vincent: That is absolutely relatable. Especially the 1 to 18, and the “if you don’t work today, you’re just going to have to do it tomorrow, so.”
Mike: That’s why I work every day. And everyone has a different style. I do not like things to build up. I don’t generally get anxiety, but if I go off the grid even for eight hours, I get a general feeling of anxiety that people are waiting for me.
Vincent: Yeah. I functionally haven’t had a weekend in years, and that’s fine. Right? Like I very much enjoy doing things this way, but that’s kind of personal preference.
Mike: Let me plug, do something you love. Scott Galloway says the opposite. If you do something you love, it doesn’t feel like you’re not taking weekends. It doesn’t. I love what I do. You love what you’re doing, right?
Vincent: Yeah. This is not what I imagined I would be doing coming out of school, but like, I don’t think I would ever trade this for anything else.
[22:20] Mike: So let’s dig into what you’re doing. Let’s dig into how you’re using AI to help you.
Vincent: Yeah. So, I’ve been in GC for a couple of different companies now. I also do have the nominal title for a couple of nonprofits, where I just kind of do a little bit of work here or there. And so my last company was a company called SkyLink, which was in the AI travel space. They were an AI travel booking and management platform. They sell to large organizations like McKinsey and BCG Travel and other places like that. And they also just got acquired by Amadeus two months ago, one and a half months ago. And so I stayed with them basically from their earliest commercial stages until the exit. And so that was really interesting, because it was basically, like, watching and helping build a company from the ground up. I also did some product work for them. Those three years were where I think AI basically became mainstream and became a necessary part of the technological stack for any company.
Mike: Were they a C-corp based out of Delaware?
Vincent: They were a C-corp based out of Delaware.
Mike: I listened to you on a podcast. If you want to be acquired, your input to someone who is building a startup, make sure you’re a C-corp based out of Delaware, correct?
Vincent: Not just acquired. Like, it is the practical expectation for most US-based VCs.
Mike: If any venture capital firm is listening to this podcast, we have no interest in being acquired. So we are not—we’re an LLC.
Vincent: Yeah. And yes, there are always exceptions.
Mike: Of course. Of course.
Vincent: You have to be exceptional to be the exception.
Mike: Of course. I hate absolutes—is there a world where everyone in our firm gets a huge deal? We have almost 50 people now; I don’t think there’s a world where everyone comes out a winner. So we have little interest in being acquired. Don’t reach out to us. Reach out to Vincent. He has a new company.
How were you using AI when you were being acquired and when you were a GC? Like what are some of real-life examples?
Vincent: Yeah, so I think I mentioned one example earlier, which is just that, like, it’s a double-check for everything. Whenever something comes in—first of all, everybody at this point should know that AI trains on your data, unless you have zero data retention turned on. And so please don’t be feeding, like, confidential trade secret company internal information into any AI tool. Bad things do happen from that. So yeah, be careful on that.
But, like, as long as you’ve got your security and privacy measures adequately prepared, AI becomes a really powerful tool to essentially do a lot of information processing that you wouldn’t be able to do at that speed. So, for example, I might get an NDA in, and in parallel to my review, I will also stick it in and just be like, “Can you just flag all the issues here?”
And the prompting is, of course, much, much longer than that. If you just say, “Will you review this for me?” it’ll give you a semi-decent, 80% of the way there prompt. If you start designing it with a couple of paragraphs of what you’re actually looking for, the context of your company and things like that, you’ll get much better results.
And then just comparing, like, what your issue list is to what the AI issue list is there, and oftentimes, it might flag, you know, one thing that maybe either you missed or something that in a different context might be important, so just be aware of it if the facts were a little bit different.
I think it also provides a great opportunity for you to make sure that what you’re putting out in terms of work product is actually usable. I think lawyers, oftentimes, when you come out of a law firm and you move into a business context for the first time, it’s very, very easy to forget you’re now talking to a bunch of people who didn’t train in law like you did. And so oftentimes, translating certain concepts for them—this is not a problem at my previous company, but it’s a problem at a lot of other companies—translating those concepts in a way that a CEO or a COO or a CFO will easily pick up on is something that AI can be really helpful for.
[25:32] Mike: What’s been your experience—as a GC, taking AI-driven work and your legal background and elucidating and illuminating it in terms that are digestible for the CEO of the company and for the B2B transactions that you’re working, with CEOs of other companies—how important are you, the human, right now at this day, into that language transfer? And how rapidly do you think AI is going to be able to do that language transfer?
Vincent: Incredibly important, still. And the reason for that is because AI doesn’t have your full context. You know all sorts of facts about the company and how to communicate with your stakeholders that AI just doesn’t, right? AI’s trained on your average generic dataset. It’s trained on, like, Reddit and Wikipedia and Stack Overflow and other things like that. And so it’ll give you, like, a semi-decent approximation of what it thinks you should do. But like I would never, ever send out something direct that AI produced for me, no matter how well I prompt it, just because there’s so many other things that I need to apply to make things more effective.
So for example, startup CEOs and company CEOs are incredibly busy. And so the sort of three-to-five-page research memo that you would produce as an associate in a law firm, they just don’t have time for that. And so, a lot of that sort of work product that you’ve been trained to do, it’s not that it functionally becomes not helpful, but it’s not enough, right? You need to distill it down into something that can be actioned on from a business point of view very, very quickly. And so that’s something where, like, as a GC in the olden days, you might look at what you produced and then kind of write down a summary and then review and go back and make sure you didn’t miss anything.
AI can produce a great first draft. I would call AI probably a great legal assistant right now. In some cases, it might even perform at like maybe a first-year associate level. Not beyond that right now, but you know, things are advancing every single day.
[27:12] Mike: I think what you’re saying—tell me if I’m wrong, but I would distill it in my layperson’s mind as—AI can’t read the room. AI doesn’t have strong EQ.
Vincent: Exactly. It’ll have EQ for what it thinks the average room is, not your room.
Mike: Is that going to go away? Can AI learn EQ? You’re talking about datasets, and my mind can’t grasp the notion of EQ ever being captured well in a dataset, but I have a feeling I’m wrong.
Vincent: There’s a lot of development in this space, actually, in the AI world, and I’m actually building in this space myself, basically building tools that give AI the personal context needed to operate at a higher fidelity. But yeah, so I would say that EQ is a combination of multiple things, right? It’s the richness of the input signal you’re giving. So when I’m on a video call with you, Mike, it’s very different from being on a phone call with you, because I can kind of see your face and your reactions and things like that. And then it’s processing all of those in a way that kind of suggests, “Based on what I know about Mike, he is probably feeling something like this,” or, “Based on what I know about my friend, they’re probably feeling something like this,” and then giving you a response that may or may not be actionable, but it’s something that you can apply to govern how you then act towards that person.
All of those things are digitizable, but for example, how are you going to get AI to read microexpressions on somebody’s face when you’re doing a pitch with them, or when you’re talking to them about a difficult matter from HR such that you might need to, like, change up what you’re saying on the spot? That is a pretty difficult thing to do. It is technically possible, right? Because I think trained computer vision will do that, feed that into a different model, and then hope that it gives me the right results. But I think we’re still pretty early in that field. I think it’s something that we will do at some point, but perhaps not yet.
Mike: Yeah, that’s fascinating. You’re basically telling me, almost, AI can make me a more likable person or be a more likable person than—I want to be more likable.
[28:50] If you were to tell Stanford, or any law school, “These are the curricular changes you need to make to incorporate AI,” what are the immediate curricular AI needs for law schools to train their students in so they can start, on day one, adding value to their firms or employers?
Vincent: So I feel like I’m going to walk into a minefield here, so please don’t take this as a blanket recommendation of what to change. I’ll just give you, kind of, things I noticed from my point of view.
Mike: Ah, you’re a lawyer. The disclaimers come.
Vincent: “It depends,” yeah.
So, AI is a necessary part of the future lawyer’s workflow. It’s a necessary part of the current lawyer’s workflow, at this point. Companies like Harvey and Legora and other ones have already embedded themselves into law firms, and they’re starting on in-house companies. Other companies like GC AI and a bunch of other ones have targeted in-house first, and they’ve already optimized their workflow for lawyers at different fields. Another company I want to mention, as well, is Gavel. That’s started by my friend Pierre and Dorna, and they’ve got a tool that plugs into Microsoft Word that essentially lets you have an AI companion while you’re drafting and reviewing and things like that.
So incredibly useful things. There’s huge time savings associated with most of those things. And there’s also efficiency, productivity, and accuracy improvements for everything.
I think that people who are practicing right now have a real advantage, because a lot of the law firm associates and partners now learned law kind of the old way, right? More of an apprenticeship model where, you know, you did the work, you did the diligence going in, or you did the doc review. And then you built up from that to drafting review memos and other things like that, to drafting agreements into reviewing agreements to negotiating agreements, and so on and so on, until you learn, you know, an entire M&A deal process or an entire lifecycle of litigation, right?
And because you did all of it, you knew how to apply that knowledge in the future. AI shortcuts a lot of that. When I want to write that memo based on the data I’ve received, I can have AI do that very, very quickly for me now. And it’ll do a pretty good job, right? It’ll do, like, a legal assistant, maybe a first-year associate’s job. And because I’ve had the experience of doing this myself before, I can review that and basically correct what the AI is doing for me and take it that last 5% so that it’s actually useful work product that won’t get us sued for malpractice.
If you’re coming in now as a law student, there’s kind of dual things that AI’s doing for you. On one hand, it’s something that you’ve grown up in, and so you are used to that efficiency gain. You’re used to being able to do research that way, things like that. But on the other hand, there’s a lot of background knowledge that’s required to find that last 5% that AI didn’t get, that you really still need to pick up, or you will be producing work product that isn’t quite there.
And we see this all the time now, right, with courts sanctioning lawyers for using AI for legal research and it making up citations or quotes and things like that. Like, never do that, please.
Mike: Right. I’ll have a video that we’ll link about a guy in court using AI to present the opening argument. We’ll link it; it’s hilarious. It’ll guarantee that no one does that in court. You’ll see the video. It’s not as good as the one we’re going to link with you on SportsCenter, but it’s close.
Vincent: So you say that nobody will do that, and yet it keeps happening, right? The first cases of this were not this year. They were last year or maybe the year before, and even just this week, I saw another article about somebody getting sanctioned for this. And so as an associate coming in from law school, developing that background knowledge is critical to making sure that you can actually grow and develop in your career.
[32:05] That being said, in law school, then, there are now two aims for every student who’s studying law. The first one is to learn the law, and that’s what traditionally people have been doing for, you know, a hundred years or more.
But the second one is making sure that you’re familiar and comfortable with the legal tooling that you’ll actually be using to practice law. And that’s where I think, incorporating AI on a—maybe not a daily basis, but maybe a weekly basis for students is extremely helpful.
It’s the same way that, if you went to law school 10 to 20 years ago, you had Westlaw, unlimited Westlaw or Bloomberg or LexiNexis, right? That was the tooling we used. AI should be viewed, I think, very similar to those. Like you should practice with it so that when you get to the law firm or to the public interest firm or whatever you want to do that you’re used to it because that’s the expectation for lawyers in the future.
Mike: Yeah, I’m glad you mentioned LexiNexis; I was on their advisory board, so thank you for adding them. So the tools are shifting. Would you hire someone out of law school if they blew your socks off in the interview with their doctrinal legal knowledge, if they had an incredible, likable personality and you needed someone who could help develop your brand, who had, like me, literally almost no experience utilizing AI. Would you hire that person or would you say—
Vincent: I probably wouldn’t, at this point.
Mike: Yeah, I thought you would say that. A hundred percent you would not hire me for your new startup.
Vincent: Oh, no, Mike, I respect you a lot. I would actually make a judgment of, “Do I think this person could pick it up reasonably quickly?” Because most people at this point are at least using, you know, Claude or ChatGPT, something like that. They’ve interacted with it in some way. It’s not too difficult to pick up basic prompting to become, you know, even 10X more efficient than a naive person is.
[33:38] Mike: What kind of questions would you ask in an interview to glean out how sophisticated that person is in using the tooling, and now the AI tooling? To your point, 15 years ago, the questions were all assessing people, their knowledge base or their ability to do whatever, and now we’re assessing their ability to incorporate AI. What would be a kind of interview question for that?
Vincent: I don’t think there’s a set industry standard, yet, for this, and I’ll tell you a couple of different approaches I’ve seen. One company I encountered recently said, “We actually have a prompting interview where we sit down with them, and we just watch them prompt.” And the reason for this is because prompting is kind of your basic interface with most chat bot AIs, and so just seeing how somebody thinks about prompting tells you a lot about their experience.
Other people will ask, “What is your AI tech stack,” right? Because some people are so sophisticated that they’ve bought Mac Minis, installed OpenClaw, you know, have an entire AI assistant in their life. And there’s other people for whom, you know, the only thing they do is they Google, and then Google has, like, an AI mode at the top, and they just look at that answer.
So even just there, there’s still sufficient differentiation that just getting the overall vibe, to borrow Gen Z terminology, of somebody’s experience with AI is probably enough at this point. Most AI right now is built to be very easily picked up by people in the domain, because it’s still very new. It hasn’t penetrated a lot of the market yet. And so GC AI, the interface looks a lot like ChatGPT, and it’s not because they copied them. It’s because that is just the most straightforward format for somebody to go from knowledge of ChatGPT to being able to interact with a product like that on a daily basis. And I’m not speaking for Cecilia here, actually, like, I haven’t actually asked her about this. This is just my personal theory on why it looks like that. But you’ll see convergence on a lot of tools on the ChatGPT and Claude interface for that reason, I think.
Mike: Interesting. You mentioned AI assistant; I actually did experiment for a month period with an AI assistant. I found it to be too manipulative, sort of in a non-binary way. The product was being too flattering to me. As people at our firm would tell you, I would rather negative feedback than pos at a feedback, because I’m never going to get better on positive feedback ever.
Vincent: Mike, I’m curious which tool this was.
Mike: Yeah. I have no idea. This is like six months ago. I’ll look it up and tell you offline. It was telling me how great my ideas were too much.
Vincent: I’m betting it was ChatGPT. ChatGPT tends to be a little bit more deferential.
Mike: I mean, it was an avatar.
[35:49] Vincent: So a lot of those companies are built off of underlying models, right? So they’re basically taking what you’re doing, processing it in some way, and then stuffing it into a foundational model. If the foundational model is a certain model, for example, that’ll treat you very differently from other foundational models.
But you can actually change that. I’ve run entire VC pitches through AI, and I’ve gotten things to go from, you know, kind of almost sycophantic to basically being like, “I want you to tear me apart like you’re a tier one VC who doesn’t really care about what I’m saying. I want to know every single thing you’re thinking in your mind and how terrible it is about the product I’m using, so that I can address all those things.” And it’ll do a really good job if you prompt it that way.
Mike: I may try that. Have you seen the movie Interstellar?
Vincent: Yes.
Mike: I knew you had, and I bet you like it.
Vincent: I’m a sci-fi person, yeah.
Mike: Yeah. Do you remember TARS? it had a humor setting. And Matthew McConaughey kept lowering TARS’ humor setting. So what you’re saying is I can go back into this AI assistant and raise the hypercritical and help me improve what I’m doing.
Vincent: Hopefully it can. It depends on how they’ve actually done their own system prompting. You’ll notice it right away if you go to Claude, for example. You can say, “Be my friend and talk as a friend would.” You can say, “Talk as if you’re an overly critical mentor or, like, my dad,” or something like that. It’ll take on that persona.
Mike: Would a good podcast be me interviewing my AI assistant about the convergence of doing things contractually for Spivey Consulting, and then me talking to the AI? Are we there yet with AI?
Vincent: You can get a pretty good approximation over a few minutes. I think anything more than maybe 5 to 10 minutes, you probably start to notice.
Mike: Okay. Interesting. Where’s this all heading, in two areas? And I know you read the transcript of my interview with Professor Nita Farahany, and she talked about cognitive extinction; we don’t have to go there.
[37:23] Where is this heading in terms of your impact on a day-to-day basis, and where is this heading, if you want to take a stab at it, on the impact of hiring out of law school? So two big questions that we could end on.
Vincent: Where this is going on a day-to-day basis, I think we’re still very, very early in AI. We have tooling from several large companies, and right now, we’re basically only looking at tools that have been trained on the generalized data set of what they found on the internet to this point. There’s a lot of different private data sets that have not been accessed.
So every law firm, for example, all of their precedent, all of their workflows, that’s all private data that large language models haven’t trained on. And so, if you give access to those sorts of private data sets to AI companies, they can build AI that is much more tailored to what you’re doing. This is actually where we’re building, for example. Like, we consider the personal data set to be the single most important data set available. So, every email you’ve ever had, every chat you’ve ever had, that all forms the basis for a lot of the knowledge that makes you you when you make decisions and act in life, for example. And so we think it’s possible to build personal AI that knows everything you know—obviously private only to you, so that you know, we’re not trading a large language model like this—but then that means that, when you build AI agents, for example, they actually act the way you would want them to act, because they know everything you know.
Mike: Vincent, am I going to be an outlier because I had a mentor who clerk for Justice O’Connor, Sandra Day O’Connor; Justice O’Connor was very private and aware that anything confidential should not be put in writing, so that was transferred to my mentor, who transferred that to me very quickly. So even in my early 20s, anything confidential, my thoughts, my angers, my frustrations or strategy, was never put in writing. Would AI take all of my emails dating back from my 20s and treat me differently because I am very buttoned up about having confidential dialogue only in verbal form?
Vincent: What do you mean by treating you differently?
Mike: Would it not know me as well? Would it not know my strategic thinking as well?
Vincent: Absolutely. Because it doesn’t have the data to be able to make that inference on you. Because remember, AI in the end is really just, “Given this, what do we think the next likely thing is going to be?” At its very base level, it’s just statistics and probability. So if the data that we put in doesn’t include a huge part of your core personality and core thoughts and memories, it’s going to give you an output that is most likely based on only what it knows about you.
Mike: When I was in my 20s, we didn’t know—other than the movie Terminator—we didn’t know what the concept of AI was. But I have hidden a strategic side of me from future AI, for better or worse. Is that for better or worse?
Vincent: I guess for better or worse would depend on what we’re optimizing for, right? If you’re optimizing for personal privacy—if you don’t want to be found out about, then what you’re doing is absolutely good. It’s something that, short of scanning in all of your notes, we would never know for 100% certainty what that side of you is. And that provides a lot of benefits to some people. They would prefer to remain private. They don’t want to be digitized, things like that.
Now, on a large population scale, AI is very, very good at inferring, right? So it can infer that there’s a certain type of person archetype out there that writes down all of their info, and these are other traits that might be associated with that kind of person. And so, if you happen to fit that archetype, it might still get most of you without getting all of you, or you could be very different from the average type of person who does that, in which case it won’t get you at all. So it really just depends on data.
[40:42] Mike: Is your company going to cannibalize your own job?
Vincent: I think that there’s always going to be room out there for creativity. I think you’ve had a bunch of podcasts out there about the after-effects of AI. If every single person uses AI and it’s the same AI, then the results coming out are going to be very similar.
Professors will talk about this. They’ll say, “I received 15 emails from all of my students, and they sound almost exactly the same. I know that they’re all going into the same AI tools and prompting them very similarly.” And so to stand out then, being creative is a really good way to do that, because it’s something that is just unexpected. And the more people sound the same, the more you stand out if you’re a little bit different.
Mike: It’s so relatable, because the market—I’ll say Reddit, but it happens on other platforms and social media—will say, “Hey, Spivey, how do I word this first sentence of the letter of continued interest?” I can’t give that because if all 800 people that read—
Vincent: Doing the exact same thing, right?
Mike: Then law schools aren’t going to like it anymore. Because, again, differentiation is the key.
It’s very relatable to me. It’s just not going to be a human. AI is actually going to have to assess—and maybe you’re saying it’s going to have to be different platforms—AI is going to have to learn to assess, “We can’t do this. We’re actually hurting our client.”
Vincent: Yeah, so, there’s an entire field out there which is basically simulations, right? As an AI thing, you can simulate things a million times, a billion times, because it’s just all processing in a computer somewhere. And so you can basically say like, show me what happens if I do the exact same thing a million times. What are all the possible outcomes? When you ask a question like that, you will occasionally get low probability events happening. Like, if you’re saying, “Simulate all the different ways I can start this letter of continued interest,” 80% of the time it’ll give you something very similar, but 10% of the time it’ll give you something completely unexpected where you’re like, “Oh, that’s actually pretty good. I can modify that and make that fit what I want to do.” And because nobody else took the time to simulate something a million times, that’s going to stand out just a little bit for you.
Mike: How would you optimize your time if you’re looking at a million different options that AI is giving you?
Vincent: I would have AI review the million things and—
Mike: Right, right, right.
Vincent: If there’s one thing AI is very good at doing, it’s pressing large amounts of data much faster than you can.
Mike: And then quantum computing is just going to be next-level that, correct?
Vincent: Yeah. Quantum computing I think is something that—that’s an entirely different conversation. We can have that conversation.
[42:45] Mike: Yeah, we’ll do a second podcast. How is hiring out of law school going to be impacted by the rapidly learning AI that is developing faster than Moore’s Law?
Vincent: If I were a law firm today, I would be picking up people that I know can go down that dual track that I mentioned earlier of both knowing the law and being able to pick up tooling very, very quickly. Because the AI tools of today are not going to be the same as the AI tools of tomorrow. Even if you’re using Harvey today, Harvey today and Harvey five years from now might be very, very different. And so being able to constantly keep up is something that is going to have to be a hallmark of your incoming associates.
I’ve actually heard law firm partners discussing like, you know, now that we have AI, do we even need to hire first year associates? And I think the answer probably is yes, still, just because you need kind of a succession plan for your firm. But I’ve also heard law firm partners talking about making this more of an apprenticeship model where they hire fewer associates but train them really, really well, give them more personalized attention, because each individual associate’s more productive. Some people have proposed things like a residency-like model, like from medicine where you train for a couple of years intensely and then come at a fully-fledged productive member of the firm as well. But in all cases, you need to be able to learn really quickly.
It’s going to be a lot like engineering where, for a long time, engineers had to pick up every new language, every new concept. Like, AI itself is a new field for engineers for the most part. And now that’s going to have to be the domain of lawyers as well, in terms of picking up new fields very quickly.
Mike: Is a fair way to put that all together, to synthesize all that is, if you’re a law school, you better be training your students on how to use these tools, and if you’re a law student, if you want one takeaway from our podcast time together is, you better be nimble and learning how to use these tools, and to an extent that you can, make it come out in the interview that you have.
[44:23] Vincent: Yeah. And I do want to give one cautionary flag, though, because I think it’s very tempting to use AI to do your work for you and then not check it. If you come out of law school having done that, and you haven’t learned anything, you will be probably the least employable person. You won’t be able to catch that last 5% that AI misses, and that makes you functionally useless to law firms and companies and public interest firms and the government and every other employer. As a lawyer, you’re being relied on for your experience and for your judgment, and you also need to develop your judgment along with developing the use of the tooling.
Mike: Yeah. A hundred percent right. People don’t pay me my hourly rate for that hour. They’re paying me for the 26 years I’ve been doing all the nuances of admissions, which AI doesn’t know. And on the admissions front, what you just said is times two, never submit an application just AI, because it’s not going to read organically over time.
Vincent: It’ll be really flat. It’ll sound like every other thing, and it will not stand out.
Mike: Yeah, we played around with it, and it’s disgusting. It’ll get better, but it’s—right now it’s disgusting. But there’s a second variable too, which I don’t think you have to confront with. If you’re an applicant, just like the C&F you had to do when you applied, there’s another C&F type question, which is, have you used AI as part of your application? If you use AI and you say no on that question, that’s a great way to never go to law school and never be able to sit for the bar.
Vincent: Yeah. Absolutely. Don’t start your legal career off with a misrepresentation.
I feel like we’re being very doomy. I think people actually should use AI on their applications. I just think that it has its place.
Mike: A hundred percent, do I want to use AI to strategize about what 25 schools to apply to, so maybe Vincent, it could be 12? A hundred percent sign off on that.
Vincent: That would have been really helpful. I would have had to pay many fewer application fees.
Mike: So, that’s a happy note to end on is, maybe we’re trending in people not having to apply to 25 law schools. Congrats on going 23 for 25. It was great to see you, Vincent.
Vincent: Thanks, Mike.
.png)
