Transcription
Hi, thank you for joining Human Capital Gains. Today, we have Marissa Parker joining us. Marissa Parker is Chief Operating Officer of Stradley Ronon Stevens and Young, where she partners with the management committee to lead firm operations across talent, IT, security, administration, and knowledge management.
"My day is generally about seeing, spotting problems, and trying to solve them."
That's awesome. Well, you know, we've had a lot of conversations. I've talked to other law firms where AI seems to be a big problem that people are looking to tackle in a meaningful way in an industry that's reluctant to for AI adoption. What kind of, kind of things have you guys seen being the biggest hurdles to AI adoption, uh, within the legal practice?
"Lawyers are skeptics by nature. I think the people in a sense is the biggest problem. It's also the biggest opportunity. We have, uh, watched both within our firm and at our competitors the, uh, kind of reaction of amazement and horror at generative AI as it became kind of a household, uh, concept, uh, back in, you know, late '22, early '23. And, you know, lawyers have a tremendous amount of ethical responsibilities that layer over everything they do. And, you know, well-trained lawyers are going to be cognizant of that, um, everything they, they do for their clients. So the initial reactions to generative AI was, 'Can we trust it?' No. 'Can we invite it into our confidential conversations?' Absolutely not. And, you know, 'Really, is it going to replace us?' Well, they could, it could never do what we could do with them. No."
And looking at those fears and kind of, uh, emotions is really where we started because lawyers, I think, when you do the Myers test across all types of, of categories of professionals, right, fall on like the ultra far end of being skeptical. It's also what makes them great at their, their craft. So recognizing that duality and trying to work with it as opposed to trying to fight it was where we started a lot of our training and thinking about how we can utilize the way that lawyers think and train, um, and then figure out places where generative AI is well-suited to be a compliment. I think that's where we're seeing in the industry and at our firm the biggest gains and the biggest opportunities for integration and adoption. Not necessarily displacement, just given the level of sophistication of the legal work we do at Stradley Ronon and many of our peer firms. I don't think there's an immediate concern about being completely replaced by the robot.
"Right?"
"It's more like, how do we, you know, put a robot next to us. We joke about this program that we use, um, called Vincent, um, as, you know, my associate Vinnie. And like, Vinnie's always available. And Vinnie's really good at three in the morning if you have to like rewrite an email. Vinnie's like on it. And, you know, there's lots of things Vinnie can't do that our associates can do that are great. But it's really nice to have Vinnie as part of our, like, associate cadre as we think about, you know, getting through efficient work processes and trying to make sure that we're bringing maximum value to time."
Yeah, it's, it's interesting, um, that you bring that up because a lot of the companies that we're talking about, I think there's two interesting points there. Is every company that moved to kind of like employee replacement first is actually dialing that back now because it's not necessarily the promise of AI can't deliver. It's that the use case of AI that they thought that they could use to replace certain divisions or certain functions wasn't really accurate. So, a lot of companies are dialing that back. And then the second thing is that, you know, there's been a couple studies that have come out recently, most recently the study out of MIT that is kind of like rocked the world, including the stock market, is that, you know, 95% of these AI pilots are failing.
And the other research is pointing to the fact that these pilots are failing because people are establishing tools first, really, and then not really bringing their people along. Right? Like we've, we've adopted a kind of mindset, skill set, tool set paradigm on how we help companies with AI implementation, and it's really in that, that order. Um, and I think kind of like what you did with your kind of Shark Tank environment, hackathon environment at the firm is fascinating, exactly what needs to happen in these industries that are super reluctant for AI adoption. So tell me a little bit more about that.
"Sure. For sure. So we, um, definitely subscribe, uh, to the people, process, technology approach. And I always say those three words are, are in that order for a reason, right? You start with people, you look at what your processes are intended to accomplish, and then you think about what technology, whether technology, and then what is correct technology and correct technical tools will improve that process for you. And so GenAI is no different in that. Um, it's just a question of what you mean when you say GenAI, right? They're kind of this like omniscient, all-purpose pool, or are they purpose-built smaller tools for specific processes? So we, we explored a number of purposeful tools and a number of kind of umbrella tools for a while and, um, ended up partnering with a company called VLAX that just recently, there was an announcement that Cleo, um, acquired VLAX for a billion dollar valuation. A fantastic company that really cares about the use cases that it's created this legal platform for. It's a legal database trained LLM, private and secure, and they're interested in rivaling the two major legal research players in the marketplace, Westlaw and LexisNexis. And they've had a stronghold for like 40 years. So, it's a pretty interesting dynamic to see somebody come in and be disruptive, and we like that."
"Yeah, absolutely."
"And beyond. So, our Head of Knowledge Management, Sarah Herabat, who, uh, was a practicing attorney and founded the knowledge management program at a global 10, uh, law firm, helped to build our knowledge management program here as well as helped be our pioneer in, in generative AI. And, and that's important because her background is about knowledge management and training, understanding you have to meet people where they are, that not all trainings are going to work for everyone in the same way. And recognizing that if you don't have a kind of cocktail-level understanding of the technology, you're not going to be able to absorb and apply it. So all of these layers need to be considered when you're introducing something that's pretty revolutionary, uh, to folks again, who were skeptical and not necessarily running towards this level of change. So Sarah designed a program with Vincent and set up a 13-session series that was focused on lawyers primarily, but we included some business professionals too, because we recognize the value of our operations having the same types of training so that we can look at problems from different lenses and bring expertise from different areas. Our folks in e-discovery, paralegals, um, legal assistants, recognizing that they have roles in the management of our clients and our client matters, and their input into where this tool can be used for their purposes is additive to the whole enterprise. And they're coming at it from a different direction. Um, so I think that that helps round out some training. The 13-step session that Sarah designed, um, with, with our counterparts at Vincent, sorry, at VLAX. Vincent is a product we use. VLAX is the company. They trained on fundamentals of what an LLM is, what the ethical overlays are for attorneys to think about in the LLM world. What use cases kind of emerged as the knee-jerk reaction of, you know, 'I want a Google, but I want a Google that also generates content.' So, what am I going to get if I use this product, and what are my expectations, and how does the product differ? And they did group work. And we really felt like the secret sauce was group work. Anybody can eventually, with enough time and resources, develop a good training program, but how that program then interacts with the people is the key."
"Right?"
"And so making sure that we built in opportunities for people to work together and problem-solve together in real time. Um, it's, it can, you know, kind of a nice flip side to why GenAI is not going to just steamroll us all, right? It's that human-centered approach to learning that I think was so key in, in being successful. So we built week over week over week. We had regular sessions with a cohort of 30 of our lawyers and professionals who invested time in as in asynchronous learning as well as with, um, group work. And then the culmination of that was a, a hackathon that was focused on three different concepts of problem-solving, um, getting, getting folks to, uh, problem-solve in a real-world scenario that could be shared and replicated with their colleagues. We're operating on a system of FOMO as well, right? If we create a fear of missing out around how to use this technology and how clients are excited about it in certain pockets, then, you know, lawyers want to, you know, lawyers are competitive and they want to say, 'Oh, well, I can do that too,' or 'I can do it better.'"
"Right?"
"We added this kind of competitive element to, to the hackathon and did a, a showcase, uh, with our participants in our cohort as well as with our colleagues at VLAX. And the lawyers came up with some solutions around pain points that they, they know their clients had experienced. In one of our co-workers, we have a former General Counsel of a major real estate company, and he was a great, you know, kind of voice in, you know, 'When I'm managing a team of lawyers and you're getting all this pressure from our business, what kinds of things do we want to be able to like press a button and...' So that program, I think, gave people a lot of confidence. And we were able to publicize it and share it, and we're getting wonderful buzz within, externally for sure, but also internally. So now we have oversubscription for our next cohort, and people are fighting to be a part of it, as opposed to training feeling like something that folks are like, 'How do I get out of it? How do I not have to do it, right? I don't need to learn this because it's just going to take time away from the things that I think are important.' That was the attitude, you know, I think we were fighting against in the early days. And we, I think, successfully shifted the attitude by demonstrating cool, useful, achievable outputs that, you know, is well-suited for."
Did you go into this program with kind of ideal use cases that you think that this tool is going to be able to like help your firm with? I guess is my first question. And then the second part to that is, did you come up with any new use cases that you'll then be able to implement, kind of like post-hackathon?
"Yes, that's, that's a great question. And I'd say both ends. We walked in knowing that legal research and document review and contract review were places where we were already using purpose-built AI, generative AI tools to test, like, capabilities. And so levels of accuracy were, were greater in spaces where you were asking for more like mathematics. But content generation still made lawyers really nervous because if you ask a generative AI tool to give you a summary of the law, that summary won't be the same every time, right? Because it's not pulling from a script, right? It's the, the, the probabilistic outputs that the, the module is building for you. So lawyers get really nervous about the idea of saying, 'Well, this is what I, you can't quote that, right? It's not a, it's not a static treaty to quote from or a case to quote from.' So generating content is, is kind of a different animal, um, for lawyers. And that, I think, is the space where we, uh, iterated and found different types of use cases. And the, the content generation is more idea generation right now rather than reliable output. Um, and I think that some of the, the refinements in the models can drive that, you know, in different directions in the future. But for now, it was much more about how do you reduce, um, reduce the time of triaging what data you have or getting up to speed on a particular matter. Those were some of the better use cases. We looked at, um, pilot of, um, reviewing contracts to eliminate, you know, concerns with the way that different clauses are termed. If you have, you know, if you, if you call, um, a certain type of clause one thing in a contract and it means something else, but the import of the clause is the same in both contracts. How do you quickly surface that? That's, that's more interesting. And I think where generative AI technology will continue to help lawyers. We talked about a program where we were trying to help standardize NDAs when companies have to, you know, complete thousands of NDAs a year, but there's slight variation based on the nature of their, you know, counterparty. And how can you standardize and allow for oversight and approval processes to flow well, um, when you also want to account for some, some level of nuance and variety? Right."
"So those are some of the areas that we explored."
"Yeah. Uh, and, you know, both of those use cases is interesting because it's, it, it utilizes kind of like LLMs in their greatest nature where it's essentially just pattern recognition. Um, and that's what they're really good at. And that's where we're seeing a lot of the most useful use cases. Um, not to be redundant, uh, across a lot of a lot of industries. And it's interesting too, because the similar reluctance that the legal profession is having, we find in kind of like heavily regulated industries everywhere. So like medical and finance, as well. When I talk to a lot of finance people, kind of like the early adopters in the finance industry are..."
"...that have been around long enough equate this to the same thing as, you know, when Excel came, came out. You had your early adopters for Excel, and then you had your people that still wanted to stick to the paper ledger. Is there any kind of like correlation in the legal profession to that, like a pinpoint in history where it's the same type of early adopters that are moving forward?"
"I think that, um, email is always something that is referenced as, you know, a mode of communication. When email first came out, lawyers were skeptical that you could maintain attorney-client privilege in the transmission of a communication. You know, there's a lot of concern. Again, a weird ethical wrinkle for the legal industry that doesn't apply necessarily elsewhere. But, you know, the ethical rules require that in order to preserve attorney-client privilege when you're providing legal advice, you have to have the hallmarks of, uh, confidentiality around it. So, if I was your lawyer counseling you and we were standing in the middle of Grand Central Station, you could easily make an argument that that's not actually privileged material because you didn't create an environment where you were protecting the material to begin with. Email was this really interesting issue in the beginning where you're like, 'Well, now we don't need to put it in the mail and wait four days to get it, but are we waving privilege by doing that?' And there was a period of time where there were some pretty, I mean, it predates me, but a period of time where there were some pretty staunch objectors to communicating with clients in any sort of real substance over email, which sounds crazy today, right?"
"But in the '90s, that was an absolute concern. And, and insurers, legal, you know, legal malpractice insurers and law firms were like, 'I don't know about the email thing.'"
"Right."
"And I feel like the pendulum swings. And the efficiency use of email and the way in which it allows us to, you know, accomplish our business are, you know, now almost taken for granted. And I think that there'll be components of generative AI that work the same way. Once people get more comfortable with, you use this system, it's the security is proven, here are the activities that you can do in it safely. Right? Email has evolved because of the world of, you know, hacking and ransomware and everything else, right? That in certain ways feel less safe than it did 15 years ago. So now we all have different types of programs where we use, you know, um, secure transfer, file share, fileshares with encrypted keys and all sorts of other things, right? So I think that that, that is a good analogous journey to where you know..."
"...what I'm seeing happening in the legal industry and the kind of guardrails you put around general AI tools."
"Yeah, I think it also speaks to that kind of like human augmentation rather than human replacement aspect of it. Like there's still, there's still a guardian kind of like watching over the output, making sure that the output's consistent and stuff like that, or should be at least, you know, I mean, I, I get a lot of emails that don't look like there was any human intervention before they're sent to me. But you can tell them. And eventually, you know, you see that moving more and more to the byways as well. You see a lot more people realizing and being able to, we've been trained to be able to tell the telltale signs of AI-generated content. More and more news organizations are pulling back to where they were before, um, with purely generated AI content, um, to kind of increase traffic and stuff because they realize that the cost-benefit analysis isn't necessarily there. Speaking of cost-benefit analysis, are you attaching kind of like any real ROI to these activities yet? Um, you know, a lot of businesses are coming to us where, obviously not in these highly regulated industries, but in industries where they've spent a lot of money on AI. People aren't really necessarily using it the way that they intended or thought that they would, and they're really not seeing any business impact. Is that a priority for you at this point, or not yet?"
"We took a little bit of a different tack than a lot of our competitors and did, um, ton of work on free tools, um, around adoption and use cases. And then we piloted a bunch of tools around and, and investigated adoption and use cases. And rather than jumping into any sort of enterprise products that are, you know, we're, we're charging eye-popping rates with unproven use cases. We often not do that."
"Yeah."
"So from a dollars and cents perspective, we haven't been that concerned about ROI because we haven't been laying out truckloads of cash to access generative AI tools. I think that that shift for us is coming because we've done these pilots and we've kind of set the foundation for how the adoption will work. And so the pricing then makes sense because you have confidence that you've built a program, whether you're licensing it from someone or you built it yourself. If you have a proven track record with adoption, um, and that kind of, you know, foundational training, then I think you're going to be in a position to say, 'Well, here's where we think we're going to deploy it. Here are the practices that we think are most, um, are most poised, in part based on the practice, but also in part based on the people.' You can't skip the people part of this, right?"
"So, if you have a group of your workforce that are interested in trying to identify efficiencies and are saying, 'You know what's really frustrating? We've done this, you know, paper editing for, you know, the last 20 years and we haven't been able to get traction in, you know, moving this to some sort of technology.' Uh, and generative AI is a perfect map for the type of work we're trying to do, then looking for a pilot and then saying, 'We would, you know, have the opportunity to, um, reduce the, the time that it takes for the processing of this type of invoice or bill or whatever it is.' Um, that becomes exciting because then your people are freed up to do much more high-level thinking. We're doing this in our IT group on a pretty regular basis because the folks who, who work in our IT group are like, the, the deeper complex projects are more exciting, but there's a certain amount of maintenance that you have to do on, say, checking integrations for one of the like 200 different applications that we have on a typical laptop, right? That's not fun work, but it's necessary work. If we can automate more of that, um, through kind of smart technology and then our people can be more involved in project development, product development, spanning things, and even in a low-no code environment. That's much more exciting than like, 'Is the application upgrade going to, you know, break our laptop?'"
"Yeah. Because you're a billable hours business. Do you have hesitation from some lawyers? This is doing that part of my job. How am I going to bill for that?"
"So we find that there is plenty of work that is not being billed, um, right now that lawyers are doing, and that's really the target for us."
"Yeah."
"It's about giving lawyers time back that is painful time, not so much saying, 'This project took 300 hours and now it'll take 100 hours.' We'll get there. And I think that that is going to be a big pricing question for the whole industry. Much more so for firms that are highly leveraged where the money flows because of the number of, of associates they have doing, you know, uh, work that could in some ways be reduced by general AI. We're a pretty unleveraged firm relative to much larger firms and Big Law. And so if you're able to say, 'Look, this is, um, client doesn't want to pay for 25 hours of research. It shouldn't they feel it shouldn't take that long.' Well, it took that long because we scoured the earth and it took 25 hours, right? But we're now only going to bill for 15 of it because that's what the client felt like the right price was. Well, if we can get to a point where we're, we're doing it in less than 15 hours, the client is happy, and we wouldn't have written off that 10 hours of time where the client felt like that wasn't the right amount to do. So we're trying to look at spaces where we're not able to bill and reduce that time because that's where you, I think, get the buy-in from. And then that flywheel continues as they realize that there's other use cases. And then you talk really about pricing with your clients and say, 'Where, where can we, um, responsibly apply this technology to save you time and money, and save us time and money? And then what else can we do for you to fill that space?' Now that we've reduced the time and money that it costs to do this one thing, we can do a lot more for you. Before, I think law firms can partner well with their clients to have this discussion and build those programs, but I don't think we're there yet in most cases with the technology."
"I think you need to prove the use case. You know, I was talking to to somebody in the medical profession that was talking about a lot of the times that they're brought in on consults, those things that generative AI will be able to most likely replace, like in the radiology space or reading an EKG. And tongue and cheek, the response was, you know, 'Those consults account for a third of my business, and I'm going to be outsourced pretty quickly.' So, I need to figure out what my business is going to look like moving forward, right? And I think that a lot of, kind of professions are looking at those kind of like really detailed and niche use cases and figuring out, kind of like, a business strategy around that. And we're finding too that the, the companies that are far more successful in bringing their people along in the process earlier are not only finding more adoption, but they're finding those niche use cases that really make a huge business impact because they bought their people in. And it's the people that are closest to the work that understands how it's going to be used most effectively. So rather than an edict coming down on, you know, how we're going to use this tool, it's really integrating those people at the front lines."
"So in the legal profession right now, um, time entry is the probably number one most hated activity. And you can't, sure, you can't bill for billing time, right? And if lawyers, you know, everybody's been saying the billable hour is going to die, 30 years later, it's not dead. Not by a long shot, right?"
"Right."
"And there are some really interesting companies, um, that are trying to solve for that pain. And if you can effectively capture time in a way that minimizes the time lawyers traditionally take to review their day, scour their devices, think through the things that they've done, um, and populate the time entries for the lawyers, um, in a way that's compliant with the way clients expect them to bill their time, you are, you were poised to be a, you know, very, uh, profitable company in the future. A company that I have in mind that we're talking to that I'm very excited to essentially work with because they're really coming from the, the viewpoint of the, of the people, the practitioner, and they understand pain points of what it means to record your day in six-minute increments and then make that a compelling story to your clients about why they should pay for that time, right? That's really what time entry is, demonstrating the value of what you've delivered. So if you're not doing it well for your clients, then you are, you know, risking not getting paid for your work. And we need to train around that and provide tools that are the most, that are facilitating that in the most inviting way. And so if we can do that, then we've given a gift to our revenue generators, and our revenue generators will generate more revenue, and everybody wins, right?"
"Absolutely. I know we have a lot of PE, VC, and startups that listen to this, listen to this recording. So there you go. There's your next idea. So as we end here, let's get out your crystal ball and talk about, not only necessarily disruption, but I like to focus on meaningful disruption. Like, where do you think we're going to see the most gains in the near term and long term? And we've talked about it a little bit, over, but if you had to, if you had to place a bet, where do you see kind of moving into the legal profession in a meaningful way in the short term and the long term?"
"Yeah, I think in the short term, it's still going to be on the fringes of actual legal services. It's going to be about, what we'll say is kind of back-office services, but around support. I think the idea that every lawyer can have a generative AI kind of a wrapper around, uh, the data and the systems that law firms use is going to be a disruption. There's a lot of work that firms will have to do to get to that point in terms of cleaning data, organizing data, and being able to reference it from a knowledge management standard. But even having a generative AI assistant that works with you to pull information out of your system or prepare documents and first drafts that feel attainable in the next few years. And that type of support work to get to that first stage, I think will, will be, you know, disruptive to the operational universe in law firms. I think future state, uh, there's, there's really, um, you know, interesting exploration around the digesting of data in large-scale document reviews. You know, that's been a, that curve has tracked along for a long time. But as the world has developed more data in all of its businesses, every time you have a transaction or a deal or a large-scale litigation, you have an amazing amount of data to digest. And there have been some very famous cases where like one little oversight on a UCCC filing has caused billions of dollars of liability at the end. And so, you know, generative, generative AI's ability to digest, to comprehend, and then provide you probabilistic outputs, and recognizing your accuracy levels are, you know, maybe in the low 80s. Like that's not going to get it done, right?"
"If that one UCCC document is really, really, really important to you, right? So there are places for GenAI in those spaces right now. But if generative AI can truly become a kind of super highly accurate analyzer of data, then that could be, you know, much bigger shifts around the way that due diligence is done for companies and contracts are managed. I think we have a lot of great takeaways out of this for not only everybody in the legal profession but anybody that's really thinking about a full-scale AI rollout in their, in their company, and really good guidelines for any digital transformation."
"It's excellent always to talk to you and hear about everything that's going on in your world."
"Yeah, it's always great. Thank you so much for being on, Marissa. Take care."