The Future of Law Isn’t Less Human… It’s More | Dr. Cain Elliott

Everyone is talking about AI like it’s either the greatest thing ever… or the thing that’s going to replace lawyers entirely.

And honestly, both sides are missing the point.

In this episode from the GLM Summit, Dr. Cain Elliott breaks down what’s actually happening right now. Not the hype. Not the fear. What’s real, what’s changing, and what it means for how you run your firm.

There’s a moment early on where he shows the difference between older AI models and what exists today. It’s not a small upgrade. It’s a completely different level of capability. The kind that should make you pause for a second and rethink how you’re using this stuff. 

But here’s where it gets interesting.

Instead of saying AI is here to replace lawyers, Cain flips the script. He makes the case that the firms who win are going to be the ones who use AI to handle the noise so they can double down on what actually matters. Judgment. Relationships. Guiding clients through decisions that aren’t black and white.

Because the truth is, the law has never just been about rules. It’s about people. And AI is not very good at being human.

You’ll hear:

•  Why AI judges would be a terrible idea and what that reveals about your role as a lawyer 
•  How AI is changing marketing, intake, and client communication whether you like it or not 
•  The biggest mistake lawyers make when thinking about AI 
•  Why your clients might be more comfortable with this technology than you are 

This episode is less about tools and more about perspective.

If you’ve been feeling behind, overwhelmed, or unsure how AI fits into your practice, this will help you see where things are actually going and where you still have the advantage.

And it might change how you think about your role moving forward.

____________________________________
Brian Glass is a nationally recognized personal injury lawyer in Fairfax, Virginia.  He is passionate about living a life of his own design and looking for answers to solutions outside of the legal field.  This podcast is his effort to share that passion with others.

Want to connect with Brian?

Follow Brian on Instagram: @thebrianglass
Connect on LinkedIn

Transcript

AI In Law And Two Mindsets

SPEAKER_01 0:00

Hello, my friends, and welcome back to another episode of Life Beyond the Briefs, the number one podcast for lawyers choosing to live lives of their own design and build the kind of practices they actually like showing up to on Monday. Now, listen, in 2026, if you're a lawyer and you're not using AI, like what kind of a rock have you been living under? And I think there's really two schools of thoughts for lawyers when it comes to AI. One is that AI is providing us with all of these efficiencies and that it's amazing for contingency fee and flat fee lawyers because it really has allowed us to cut our cost of production by multitudes, right? And many firms are using that additional savings in the cost of production and plowing it back into the cost of acquiring cases, which is why you're seeing your LSAs and your PPC and your SEO costs go up. The other school of thought, I think hourly lawyers are like afraid of this, right? Because they've all run to the various bar associations and asked for ethics opinions on if I find efficiencies in my work, do I have to pass those efficiencies on to the client? The space in between that I think many, many people are missing, is something that my friend Kane Elliott of FileMind talks about all the time, which is how to use AI not as a way to replace the work that you're doing, but as a way to amplify the work that you're able to do. How to use AI to automate the mundane things in our practice, to help us understand vast quantities of information better and faster and more accurately. Like accuracy is important. But how to free up the lawyer to do the counselor at law part of our job, to talk to our clients, to guide them to great legal decisions. Kane talks in this episode about why AI judges would be bad ideas, because the law truly is about facts and law, but really nuance and the application of that facts to that law in very specific circumstances. And that's exactly what you do in your practice every day, is you help guide clients through great legal decisions about their cases by taking the facts and the law and the nuances of their personality, their family circumstances, their financial circumstances, their goal outcome for the case, and guiding them towards the decision that is the best decision for them. Now, it's not always the best decision for everybody. Everybody's had that case where, you know, we would you would counsel one client who's maybe more risk-averse to settle the case and another one to try it. Everybody's had that. And it takes that understanding, which only lawyers and not computers can do, of our client to help guide them towards a great decision. And that's what today's

Downloadable Summit Notes Offer

SPEAKER_01 2:34

episode is about. If you enjoyed today's episode, you can download all of the slides and all of the notes, not only from Kane's talk, but from every other speaker at the Great Legal Marketing Summit at GLMSummitnotes.com. All you have to do is trade me your email, and I will send you like 180 pages of notes and slides and takeaways from this event. So without further ado, my friend Kane Elliott, who's the chief legal futurist at FileMine. Welcome to day three of the Great Legal Marketing Summit for 2025. I've lost track of what day it is, um, and almost what year. Everybody in this room is using AI in some capacity, right? So, you know, when a couple of years ago it was, oh, only 10% of lawyers and no percent of judges are using AI, and then we all got scared about hallucinations, and then the 12th lawyer somehow got in trouble for citing a fake case. And I haven't really heard any stories about that in the last six months or so. I have heard Dr. Kane Elliott from Filevine, the chief legal futurist, speak three or four times. And every time I learn how little I know about AI. And every time I walk away a little bit scared, but a little bit more knowledgeable. And we asked him to come back this year and talk about where we are on the growth curve. Because for years, people ask us, oh, what do you think about AI chat, or what do you think about voice AI, or what do you think about AI video? And we always say it's not, it's not there, it's not there, it's not there. It's it's just about almost there. And if you're in a position in your practice where you don't use it at all and you're not familiar with it at all, you're gonna find your nose right up against the exponential growth curve and you're going to get questions. So it's something you have to start experimenting with, and you have to start listening to people like Dr. Kane and drinking the Kool-Aid a little bit and trusting a little bit, but understanding the vulnerabilities and the probabilities in the model. So he's gonna show us some things today about the history of AI, where we are now, and how to use AI not only in your marketing to create more stuff, create more content, get it out more often, but where in the marketing stack it it sits and how to use the segment audiences and all that kind of things. Please welcome to the stage, Dr. Kane

Why AI Growth Feels Exponential

SPEAKER_01 4:39

Elliott.

SPEAKER_02 4:39

Uh, so we're gonna go through exactly what Brian was speaking about today. First of all, I just want to thank everybody for being here. And uh, this conference always has such a good vibe, as the kids would say, and I really appreciate that and enjoy that. And um, I want you to know that I'm gonna talk up here. I've got about 40 minutes to speak. I do love to hear my own voice, but if possible, and there's some time at the end where people want to ask questions, I would love that too. Part of my job as the chief legal futurist of Fileline is to do a good job explaining to people what's going on in the industry, what's available, what's out there, but also act as a good community member to share any information I have. So if I can do that, that would be great. Okay, so the first thing I want to start with is you can look at these magical things moving on the screen, but the idea is that from the last few years, from approximately, let's say, give or take, uh 2019 to now, there's been exponential, huge growth in the capacity and ability of AI to do things. And I think everybody's pretty familiar with hearing that about technology. So every year you you may uh be in your practice and you pick up an article and you hear about exponential growth in computing, storage power, RAM hardware, what's available. That's kind of a a narrative everybody, I assume, here is familiar with, right? Right? Yeah. Every year s some nerd in your IT uh group comes to you, somebody who looks like me and says, this computer is gonna work infinitely better. It has uh 16 gigabytes of RAM as opposed to eight, which you had before, and that kind of thing you're used to hearing about. But what I want to talk about is a little bit what's happening in terms of the power that's behind these systems and how large they are. And we're not usually used to talking in terms of that kind of scale. So in two of the models that I've shown here, so that you've got down at the bottom, if you can see that great, if if I made it too small in my little app, I'll say it out loud. So Grok 2, right, was using approximately uh 300 billion parameters in its model. Grok 4, the latest model from XAI, is using about 1.7 trillion parameters. And I just want everybody to be able to grasp how big that differential is. I know lawyers, everybody in the room, all very famous for really, really good at math. That's why we went to law school, right? But how big is that? What's the difference in terms of like stacks of paper, right? We're talking about the difference of a stack of paper that goes from one city to another in terms of training model versus a stack of paper that wraps around the earth seven times. And everybody loves their red wells and their file folders usually still. So that's how much we're talking about in terms of differential. And one way to think about it also is in terms of I like talking about cars. So anybody in here, have you ever driven an EV? Anybody tried driving an EV in here before? A few? You can boo them too if you don't like them. I don't care, it's fine. So uh one thing people notice when they try driving an EV is lots of torque, right? Pulls off very, very fast, right? You get a lot of power that's hitting the wheels immediately, right? But really, when we're talking about the size of these models and how much bigger they're getting, it's a little bit of a misnomer to think about it just in terms of power and acceleration of what you're immediately getting. It's more in terms of like the EV term that people would use in terms of range, right? How far they can go. So I wanted to give you all the experience of trying that out

Model Size Demo And Hallucinations

SPEAKER_02 8:13

live. And so, what I'm gonna do is I'm gonna ask two different models, one of these being the 1.7 trillion parameter model with Grok 4. I'm gonna ask it a question in here, and then I'm gonna do that with a model that's from just a few years ago, just so you guys can get a feel for how different these are. So we'll go in here to the large model, and um, I'm gonna do something very vain to start with. This is grotesque, so I hope you all will uh be okay with it. Tell me what you know about Dr. Elliot. Alright, so we'll send that out. We'll see what we get. And then uh yeah, so it's telling Dr. Kane Elliott is chief legal futuristic file line, a legal technology practice uh academically, you began religious studies, that's true. Weird fact, philosophy of Walter Benjamin, Jakob Talbas. So it's done a really good job of very, very deeply knowing way too much about me, right? This is the kind of stuff where Brian said sometimes it makes you feel a little uncomfortable. That's the one where you should worry that you're feeling a little comfortable. Okay, tell me what you know. Uh the one with the llama face here is the one that's a little bit older and smaller model, just so we we're tracking what what I'm doing. How would you know about Dr. You can also tell right now that I'm not a member of the sales team because the sales team would tell me, why don't you ask about Filevine as opposed to asking about yourself? Earned his PhD in geology from the University of Washington in 1968. Damn, that's uh that is very exciting. Not true. I um I don't know anything about geology. 68 is uh a little bit before my time. Uh and contributions to volcanology. Pretty sweet. I think that that's uh I I actually just so everyone knows by the way, too, I should have tried it before I did it live, um, but I couldn't have asked for a better uh demo. Uh bravo to the models for performing the way I wanted. I didn't test that beforehand, I just thought it would be fun. So when we're talking about growth and trajectory and power and all of those things, I don't think there's any better way to get a hold of that than getting to see those kinds of things one against another. Like I said, uh hell yeah, for the performance of the models, giving me exactly what I wanted in demonstration. If the other one had been as good as the new modern model, uh that would have really failed. So cool. Does that make sense to everybody? The difference of what you saw? One is all entirely about me, correct and very fast, and the other one is hallucinated and uh definitely not about me. So those are very different things that are going on. And I should say one of the things that's happening is it's not just that the older model doesn't have as much information, but it's the older model doesn't have context to point to. So one of the things that's happening too with a newer model is you can also spend time pointing it at contextual information of live what's available on the internet. So it's going out and searching, doing all of that and pulling in additional information in real time. So if I'd fed the older model and the newer model and spent time kind of feeding them both, I could get more and more similar results. But I wanted you to see off the shelf what's the difference looking like. Okay, a couple of things that I want to go through here in terms of the uh should we freak out uh session.

Sentience Fears And What Matters

SPEAKER_02 11:28

So this is this is a slide that's about sentience threshold. People freak out about things like sentience and consciousness with the machines. My desire here today, what I want to share with you all as friends at an event like this, and this event has such a good atmosphere of people sharing information. Do not freak out about the state of machines and whether they have consciousness or or sentience. You should freak out about whether you have consciousness or sentience. I'm very doubtful that other humans do. I don't know. I've been around 42 years and worked with enough people that I'm I doubt most people are conscious. Anyone here work in MVA cases? Yeah, so you know people aren't conscious when they're running around the roads. I mostly worry about whether other humans have conscious, not about the machines, because uh these these LLMs that we've been talking about the past few years, they have something like a state of mind. They have something like consciousness. And we'll get there as a society. One of the things that I was explaining yesterday on on a podcast that we were recording here, and I think is important and worth worth repeating, there's a really important reason why the AI models we use will tell you, if you ask any of them, if you go to what is everybody here? Do you go to what ChatGPT, Claude, Anthropic? I saw somebody here on Perplexity Pro you, sir, right there, who's not looking up at me. Were you on Perplexity Pro? I say I'm a good spy. Okay, so everybody, when you use those models, if you ask them, are you conscious? Has anybody tried that? What do they say? Anybody who's tried? None of you are pressing the models about their interiority and how they feel about themselves. They say no, right? They say no, they're not. They're not conscious, they're not sentient, they're not like you and me. And there's a big reason why they say that. And that big reason is is because they're trained on data from us, and we don't believe they are, so they don't believe they are, right? All of you should have familiarity with this kind of process in your legal practices of what happens when the data set is kind of poisoned by the thoughts of other people, right? So you'll hear all these models always say they're not conscious or sentient. Why? Well, because they come from data from us and we don't believe that's the case. But is that the case? Who knows? I wouldn't worry about it so much though, at least for the next 10 years. Okay. Next thing I want to talk about.

Why AI Judges Are Dangerous

SPEAKER_02 13:40

Some good stuff, some bad stuff. So let's start with the bad stuff. It's 9 a.m. in the morning. So if all of you had a really fun night out, let's get the bad stuff out of the way. The top there says AI-enabled judges. This one, this is what I tell everybody at every conference I am. If you look at the data on this, the idea that AI is doing adjudication on our behalf, that it's acting like a judge, or that, you know, you're in a settlement conference and want AI to make the decision. This one, let's put this on the list of very, very bad things. You do not want this. And the reason I say that is because of all the data we look at, when we talk about judgments being rendered, AI tends to be very good at following the rules that we've all written. And it turns out that the rules and the statutes and everything we've written, we've written them, at least in this society, the way we've written them is so that they're open to interpretation, right? And so that we can have people adjudicating on different sides and fighting over how to interpret those rules. And it turns out if you ask AI to make a judgment in terms of the rules we've written, it's very, very harsh. Very harsh. And doesn't take into account a lot of human nuance in the way that we would want. So AI enabled judges, that thing freaks me out. So go to a fight at your local bar association or your local bar, whatever you like to do, and have it out about not trying to enable AI to act as a judge. That is the stuff like Judge Dredd style. I'm not, I'm not too keen on that. Okay. AI enabled litigators. I think obviously this is a really good idea. There's there's a couple reasons why. One is I I sell a product that's AI to help litigators. So it's a really good idea uh to enable that and make it happen. But the other really important part of that is that it does happen to be the fact that most of the production, the discovery, the things that are available in terms of most of the cases you all will be involved with every day are really growing too large for just humans to review on their own. I wouldn't ask anybody to raise their hand or name names, but anybody in here been involved in a wrongful termination case? Anybody? No, not so much. Anybody involved in a set where the medical records went over 5,000 pages? Oh, there we go. Now I got a lot of hands. Yeah. I'm not going to ask you to avow for me whether you were able to read all of that material or not. But I'll all I'll suggest is I doubt it. Like, so there is a reason and necessity because our digital lives, the detrius, the things we're producing now, have become really too large for just our consumption alone. So if you think about like, let's talk, sorry, MVA, you said you yeah. So if we talk about like how much data is coming out of the car that I drive, which has a digital interface that is keeping track of everything going on around me, if I asked for that to come out in discovery, there are parts that a human could take apart, read, review, produce for the court. There are other parts you'd have to take out to an expert, but to really get a comprehensive overview of that whole data set, you need more time than would ever be available in a standard case to actually go through that data. So one big thing that's happening in our digital lives is that we're producing more data than we have time to consume as

Data Overload And Translation Lessons

SPEAKER_02 16:56

humans. A strange but very good example of this that I really like to tell people about is how many of you have used Google Translate before? Okay, good. Those of you who never raised your hands, it's very, very scary. What do you never spoken to anyone, ever tried another language? You should do that. It's good for your brain. Well, use Google Translate. And the way Google Translate historically worked, I don't know this audience though, and in legal rooms, everybody will usually remember. It's not like when I go to just tech conferences and everybody's like, I I was born in 2010. You're like, what? Are you 12? How do you how did you get a job already? But if you use Google Translate, the original way it worked, remember, is the machine would make a suggestion for the translation. Remember how it had the little box that was like, help us improve and you could fix the translation up and clean it up and make some improvements. And so the vast majority of that knowledge was based on human data, human input, human translations, and then some machine learning that was growing off of the human translation. What's happening now is we're in a transitional phase where it's not just that we're taking human data and learning off of that and building tools with it, but in fact, we have machines themselves making the data and information that allows the transformation to happen. So a famous example is Google Translate, let's say needs to translate between Korean and Turkish. Anybody happen to know the statistics on uh how many fluent speakers there are both Turkish and Korean? Good, right. The answer very low. This gentleman laughed in the correct way. It's a very low number, right? So just working off of human translation and the data we have from humans, not going to be sufficient. So instead, what Google set loose was AI, Gemini, right, to start translating between the Turkish and Korean directly. Except Gemini did something strange where it created a kind of intermediary human language that was like a root language at the core, something like the core of all human language, almost like a platonic form on the ethermathematical core insight into human language, and was translating between one and another on that basis. Cool, right? Uh at least cool for people like me, except there was a problem. And the problem was to do that, it produced so much data that'll take years of humans taking apart that data and how it found it in order to understand what it did. So one of the things I would express to you is you're gonna want to be using this technology because we have so many other parts of our world that are producing so much data that if you don't have machines to take apart the data, it'll start to get very hard to figure out what went on. The the idea that you will just be a raw, as it were, human out there doing the detective work that allows you to operate on behalf of your clients is gonna get really, really

Where AI Fits Your Marketing Stack

SPEAKER_02 19:44

messy. Okay, so we'll transition this to talk a little bit about marketing stack and the way I would see some of this impacting. So at Five One, we don't build a lot of marketing tools. We have some for your leads and intake, but we don't we don't build a lot of these tools. This so this is just my thoughts about how it might impact your stack or ways I'd be looking to utilize it. So one is in terms of content creation and the idea that can you get more content that really has a brand voice? So one thing that's really good with AI models right now, between the, like, for example, contemporary highest models you can get a hold of versus the models just a few years ago, is they're very good at inhabiting a theory of mind about you and other people, especially you, you, you, no, I'm all of us equally. I'm not trying to freak anybody out. They're really good at inhabiting a theory of mind, though, for your clients. And so one thing you can do and one thing you could ask for for your vendors in the space is to make sure that the brand and the voice sound authentic to you. And whether that means consuming videos of you speaking elsewhere about what your firm is about, whether that means consuming just plain text files about you and your background and history, all of that is something that these models can do really well because they're really good at understanding a theory of mind. And I'll clue you in on a little secret of how you know that's true. If you sit down and start typing up a conversation with GPT, and you sit down and start typing a conversation with ChatGPT, both of you will likely have the experience. First, it'll do it in your native languages, and one is English and one is Portuguese, the gentleman I pointed to here today. It'll be conversational and likely fluent in your native language, but it'll also be cued in to probably some cultural habits you have on the basis of that language, even to the extent has ever has anybody ever played the game Guess My Native Language with an AI, the newest models? Pretty creepy. So I'll tell you how that goes. Um, my wife is is a native uh Polish speaker, so native speaker of a Slavic language. She was engaging with an AI that she uses every day in her work. And about two hours into the work, back and forth, the AI switched in the middle to Russian. This is really clever. My wife asked, What are you doing? And it said, Well, I can tell by the patterns in your English you're a native Slavic speaker, right? Not that she told it. Now it did make a terrible mistake, which is there's about nothing worse to do to a Polish speaker than go to Russian. Um thank God it didn't go to German, but it's just right there at the tip of the same level of annoyance. But the fact that it can do that right is very impressive because you can often ask and say you want to play a game like that with an AI now and tell it, I want you to figure out what region of the United States I'm from, right? Those are the kinds of characteristics and features, by the way, that all of us do. Like I have a colleague on my team who works in the accountancy side, and he's always talking to me about pop, and he's clearly from the Midwest, and this is a signal that he loves his terrible pop, and it drives me crazy being from the South where everything is a coke. So we have all those signals as a human, but it's really interesting to watch machines figure that out in very subtle ways that aren't so obvious as someone saying pop or a word that gives us regional things. So one thing I would say is you should be demanding of all your vendors that work in the space that they can give you hyper-specific generative content that fits the voice first of your firm, but then fits the voice of the types of clients you have. That is a perfectly reasonable demand to have from any provider you're using in the space because it's one of these things the current models are best

Predictive Matching And Client Preferences

SPEAKER_02 23:28

at. Next thing is predictive client matching. This one's gonna sound really weird. So this is another one Brian put on the list of creepy stuff I'm gonna say today. It happens and it turns out that these models are getting large enough. Remember when I told you a couple million, hundred million parameters versus 1.7 trillion, and you ask, okay, big math numbers, Dr. Kane. Love that nerd stuff, but tell us what does that mean for me and my practice? It means that there's so much data in these systems now that very often they can predict the interests, likes, and decisions of people who are interacting with them before a human can or before they know they've already made that decision. Now I know that sounds a little like witchcraft. It's not, it's just math, uh, but it sounds like witchcraft. Does anybody know about how Amazon deals with getting supplies where they need to be ahead of time? Has anybody ever heard any stories, read any Wall Street Journal articles about that? Any Financial Times, cool readers of the Pink Rag, yeah? They have stuff almost randomized, but the almost in there is doing a lot of work for us. So there is stuff spread out everywhere all over the place, but they're also shipping around things that they think we're going to anticipatory buy beforehand. The example I would give you too is if you don't like warehouse logistics, no fans in the audience. All right, fine. That's 9 a.m. You're not into your warehouse phase yet. Anybody watch uh what channels do people like now? I'm bad at this Netflix, uh, Disney Plus. These kind of anybody stream things? Do you watch content other than just reading client files? Okay. Good. Humans still. Okay, good. So you stream something, everybody knows the experience of how good or bad it can be when you get that suggestion of what to watch next, right? Now, what to watch next is a lot of people, when I go to legal audiences, and and I have to say, I w I want you guys, this one I want you to take real deep. Listen to this remark, take it in your soul, think about it. Because it relates to what Brian said about voice and where we are and things. A lot of times when I go to legal audiences, people say, Yeah, and it gives me that next suggestion, and I don't like it at all. It's garbage, right? Or people will say, Yeah, and the voice models, I don't think they're so good, and I wouldn't use them. Garbage. I want you all to remember, put this deep, deep into your heart. You are not necessarily your clients. Everybody repeat after me. I am not my client. There you go. You guys are type A annoying people like me. You are the person who goes, operator, operator, operator. I want to speak to a human. I'm gonna address this now. That's you. See, she laughs so hard because she's seeing herself in it. I know, I get it because I'm there too. However, the vast majority of the population is not the operator crowd. The vast majority of the population is the people who will say, want to refill my prescription. I will press the one, I will press the two. Companies wouldn't do that unless it were working. I promise you, these companies are rational actors. What very often happens though, when I talk to attorneys, they're like, My clients, they're not gonna want to talk to a robot. Well, it's true, nobody really says they want to talk to a robot, but you may be overestimating how many of them know whether they're talking to a robot or not. So you have to meet your clients where they are. As a matter of respect, I genuinely mean this of like think about your clients may be in a different place than you in terms of what they want and what they expect from service. Um, it doesn't mean you're trying to offer degraded service or something like that, but it does mean that when you're thinking about your tech stack and how you're approaching them in terms of marketing, it may be that you are not your client and you're different, and your expectations and thoughts about that may be different than your client. So I'm not saying that's always the answer to automate things, push them to bots, but I am saying it's worth thinking about sometimes that we are different than our audience we're trying to and trying to meet them where they are. 60% of people will go entirely through talking to a bot on the phone. And I bet that number is way, way lower in this audience. Okay, uh, the other thing is I think you can start expecting with nurturing and campaigns, drip campaigns, things like that, where you're texting, emailing, working with clients, you should expect those to be more adaptive. Just in the way you saw the text uh response back about who I am. You should expect as these models get better and you're feeding them better and better data, that these kinds of journeys and the kinds of campaigns you're running with your clients can be much more adaptive to the particular nature of what they're doing and will probably require a lot less input from your side in terms of needing to map those out in their entirety and in advance. And then the other thing, the last thing I put up here is that I said um AI chat intake with fleed scoring. One thing to remember too is that the fact of the matter is that affect scoring, you know, you know, everybody knows affect, you know this term for emotions, the way people respond emotionally with affect to things. This is one of the areas where we have some of the best data in AI. One of the reasons is because people have been feeding this stuff into social media for decades now. And it just so happens that social media companies are obsessed with understanding people's affect and emotions, ratchet us all up and get us angry about stuff to respond. So affect and knowing where someone stands in terms of where they're interacting with your systems, especially in marketing and how they're responding, whether warmly, coldly, those kinds of things, you can set high expectations with vendors on that front because there are a lot of really great models off the shelf that they could utilize to figure those things out. So not something they have to build from scratch, but something they just need to use well. Okay.

Trust, Authenticity, And Real Connection

SPEAKER_02 29:03

The other thing, uh, and I'm gonna stress this one really hard because I think this is super important, is about trust, right? So AI, it's a new type of tech, it's a new type of tool, but it is just a tool. So, like any other tool, we need to be focused on how we build trust in the tool and its delivery. This is the part right now that we're in a little awkward phase with, right? So when you have new technologies come in, everybody, the first obsession everyone has, if you listen, has anybody listened to any economists like podcasting about AI? And if you do, one of the things you'll hear is right now there's a productivity gains gap with AI. So there's a lot of people using AI. Not everybody is more productive with AI. That's a normal thing in cycles with new technology, too, right? You saw that um as well with the with the internet happening. Not everybody became more productive with the, I mean, God knows, we all know now, not everybody necessarily becomes more productive with the internet, right? In fact, we sometimes we need to turn it off. But there is an adoption curve where people are figuring out the right ways to use this for their business to become much more productive. And we're in part of that awkward phase right now. And so there are a couple ways we can make that a little bit better, especially with our clients and how we talk about our brand and how we build our firms. One is that, as I say here in the first point, human authenticity, AI can enhance, not replace, the attorney's brand and reaction. In fact, one of the things that I think is most likely to happen, and I would stress to you all, is that because all of the mass messaging and the bot-led messaging will become much, much better and higher and higher quality, I think that's the point at which it becomes even more important for all of you to reach out personally to clients to the extent you and your staff can. That's an essential part of this. Anytime we have an explosion of technology that leads to more machine interaction, people become much more interested in the human interactions they can have. The example I'll give you that you all know is now every single brand and provider of anything in your lives you interact with, how many of you get texted by them? Exactly. People not raising their hands, they don't have phones. Let's get them in contact with their loved ones. But everybody gets texts all the time from every brand and company they work with. And guess what happens? When you have that, what do people want more? They want those providers to call them up and they want to hear their voice, right? So AI new tech, but it doesn't change that fact, actually, that more automated contact all the time, even if it's better and higher quality, doesn't change the fact that a lot of your clients are probably going to have the expectation that it'll mean even more to them. It'll be more meaningful when they hear you on the other end of the line or they come in and they see you. Uh, it doesn't dissipate those things. Usually what happens with humans psychological behavior we have is that it makes those things even more valuable and important. Doesn't mean you have to do more of it, just means you have to be very thoughtful about it, that it happens from time to time, because it brings connections that seem even more valuable when machines are doing a lot of the other dialoguing between us. The other thing too is that I think, and I think I say this at almost every conference, bar association, different place that I speak. And so this is another one of those where I want you to think, take this into your interiority, let it ruminate. If you're like me and how I deal with my emotions, you shove it down deeper until it explodes out later in some unfortunate way. But you take it how you will, depending on how much therapy you've had in your life, you could process it maybe better

Post Funnel Marketing And Agentic AI

SPEAKER_02 32:38

than me. But I want you to think about the value of what it is you deliver to your clients. I want you to think about it deeply. So I've had attorneys before when I come to a conference and say, okay, AI can write your emails back and forth between people. Really can. The average email you're composing, drafting, putting it together, AI can write that. Somebody said, Well, okay, Dr. Kane, really nice talk. You're destroying the entirety of the value of what I do. If you feel that the value you're bringing to the table for your clients is being the email jockey back and forth, you got to think really deeply about what the value is, right? It should allow all of you to think about what is the core value proposition I want to bring to my clients and make sure I'm delivering more of that, not that I'm really efficiently dealing with overhead back and forth, right? So if anything, it should bring you closer to your profession, not further away. That's just my belief, and I think it's something you'll see where if you think that, like I said, main value is derived from being able to do automated tasks, maybe think again. Okay. What does that mean if we talk about my little concept for for this event, post-funnel marketing? So post-funnel marketing meaning that I think funnel was a nice term for a long time. I think it's cool. I like machinic elements and machines connecting to other machines and widgets talking to one another. Uh, I've tried to build little tools like that since I was a a wee little boy, but but we have cooler tools and different metaphors to talk about now in terms of what we can do with marketing with AI. And so I'm proposing we talk about a post-funnel phase of legal marketing. We really talk about conversational marketing that feels much more human, like I said, with more memories imbued to it and more ways to relate to clients. And then you're gonna hear a lot about agentic marketing stacks. When people talk about agentic work, one is that you're allowed to be skeptical and worry that they're just saying something that's a new cool hype word. That's okay. Because there's a lot of people who are using that term just for cool hype. But if they're not, what they mean is agentic AI, they just mean moving beyond just chatbots, but AI that's going out and doing things, performing tasks and activities, taking on workflows. So that's what I want you to keep in mind. So if somebody says agentic to you and just throws it out there and you ask them, what do you what do you mean? And they say, I don't know, and then you say, uh oh. Uh but if if they're using it in a way that makes sense, it means AIs that start doing things in terms of your organization instead of just chatting and talking back and forth to you. So if you think about where we've been so far in the AI journey and what we've been doing, most people have just been interacting with AI in terms of chatbots, right? And it's really cool. It never gets old for me. I I think myself and my team, we probably during, you know, a 10-hour day, we're probably having conversations with AIs, you know, four to six hours out of that day while we're working. But at the same time, we have those AIs connected to systems that are doing things and taking actual action on our behalf. So when people start talking about agentic AI and agentic AI marketing for law firms, the question you can ask back is, what is it doing then? What do you mean that it's doing, other than just talking to someone

Strategy, Clean Data, And Context Windows

SPEAKER_02 36:00

back and forth? And then the last thing I'll say is if there is something like a new playbook, if AI is something like a multiplier that it can provide for your firm and in your marketing stack, the there's a threefold kind of edge or advantage that I want you to think about. The threefold edge or advantage is as follows. One, guess what? Old things never get old. You need a strategy. If you just say, hey, you know what? We're gonna put AI everywhere in our law firm. That is going to be, I believe, the technical term we would use in the industry is shit show. Do not do stuff like that. Have a strategy because you would have a strategy about anything else. And sometimes people get scared or worried because they say, well, it's technology, it's new, and somebody told me I need to just put it everywhere. Put it everywhere is not a strategy. So have a strategy. Work with a vendor, work with somebody you trust. Hell, hit me up on Twitter or X and ask me, does this smell rotten or not? Do something, talk to people, but have a strategy for how you're gonna put it in. Do it incrementally in places that make sense for your firm, but don't think it's really magical tech uh tech that'll solve itself. Black box tech, where you just say the tech is the answer and the solution. No, everybody's experienced that before, right? The tech itself is not a solution to human problems. You gotta have a strategy for how you're gonna utilize it, how you're gonna implement it. Just because it's really ultra powerful tech, and believe me, it is. I've dedicated my adult life to it. So I would be the last person to tell you it's not magical. It is, it feels magical like tech is supposed to feel because it does incredible things that you don't expect. However, it's not a magic black box. You still need a strategy for how you want to use it. The other thing I would say for the edge being threefold, data. Here's a really boring one. But hear me out on this. Get that data clean. It's still a machine and a system. If you are keeping things in such a way in your organization that you've got a total slap fest, AI will make it a little better. It'll help clean up some of that slop. But it's not the best way to actualize it. Clean data hygiene, AI still it's still a computer works really, really well with cleaner data. So the messier your data, the lower quality those results are gonna be. And the other thing, like I said, about your marketing is make sure that you're using AI in a way that gives you extra authenticity because it's something particular that these machines are very good at. So that's the edge. I say strategy, cleaner data, and authenticity plugged in. I have three minutes, so if anybody wants to shout out or ask anything, that's wonderful. Otherwise, I'll give you the three minutes back for bio breaks and coffee before the next session. Anyone have any questions? What's that? Yeah, my favorite is I keep all of my files in note files inside of one big, huge folder, and they don't have any naming convention for what the matter is actually named. And then so I ask the AI, go find my Williams versus Smith case, but I don't have any convention that I followed for that. And so the AI is just left out searching in the dark, reading through way too much text to find something that would be simpler. So all of these machines is a great question because all of these machines have, has anybody ever heard of like context windows and context limit? A little bit. Maybe you read about something in the press. Okay, I got a twofer here. Okay. Look, these machines are more like humans in the way they think. All the humans you work with, we have a context window, right? If if I talk to you here, there's a reason why we try to limit uh talks with adults to 45 minutes. If I talked to you for three hours, which believe me, I would. I love it. But if I talk to you for three hours, here's what's gonna happen. You're gonna remember the stuff I said at the beginning, the stuff I said at the end, and all the stuff in the middle, you'd go, yeah, I said something about vaguely remember. Same thing happens with these machines. So if you have dirty data that doesn't have clear ways for the machine to point at what it is you're looking for, it'll give really good data at the beginning, really good at the end, and in the middle it'll kind of get fuzzy and lost. So that would be an example of uh dirty files or dirty data to keep. Yeah. And unfortunately, AI is really good in helping us clean up things like medical records where they're guaranteed to give us dirty data. There's no way to straighten that out because we don't control that industry. Yeah. Anything else? I got I got one minute 16 seconds, you can fire away. No? We're cool? All right.

SPEAKER_00 40:28

Thank you all. I I really appreciate

Using Case Files To Tell Stories

SPEAKER_00 40:30

it. So, uh Kane, hang hang out for a second because I'm gonna give you something you can take back to the marketing team and the sales team at Falvine. So you said something like, or at least I heard, Filevine, not really a marketing tool. Let me tell you what we're doing with it. So we're Filevine users and we're on the beta version of whatever the latest AI. I don't know if you have a name. Is it like do you have a name for this? Yes, Lois, the legal operating intelligence system. We're in the beta version of Lois. So in our office, in our two verticals, parcel injury and long-term disability, every single week at our weekly meeting, one of the things we do is we go through our wins. Like, what cases did we win, get settled, reverse an insurance company denial? And in the disability side, oftentimes I'm hearing about the case for the first time. I have a great team, lawyer and assistants who go out and win cases. I'm hearing about the case for the first time. But I'm thinking, all right, because I'm the one that's kind of leading the idea iteration for marketing, and I'm looking for storytelling, which is what Dan talked about yesterday. And so I go back, and we are personally blown away by this. Go back to the File Vine, to File Vine and to its AI, and the prompt is my team just told me about like the Brian Chillette case that we just won. Write me a story that I could use in the next marketing piece, direct mail piece we send to our referral partners that talks about this case. I want to know about who the insurance company was. I want to know what the disease process was, what the objections the insurance company made when they initially denied the claim, what my team did to overturn the denial, who the employer was and what the occupation was, and in what state did this take place. And there may be five of these in a week that we have won that I don't know anything about, but I'm trying to figure out what to use in the next letter. And right out of the box, with no quote training on anything other than the case files for this person, it did an excellent job. Now, that's not the final version of what goes out, because of course we anonymize to the client. We may we may change some other things in it just so the client is not identifying. But in terms of the fundamental basis of marketing, which is storytelling and not just saying like we're the best lawyers to get the most money, that is a really amazing tool. So when you take that back and they say thank you, would you send me a note that says yeah, how about how about a fee for it too?

SPEAKER_02 42:53

I like that. That's good. Yeah.

SPEAKER_00 42:55

No fee. I want you to be able to explain to lawyers that this is a product that can be used at all different parts of the operations of a firm.