From Outdated Law Firm SOPs to AI Prompt Libraries and Internal Chatbots

0:00 Introduction – Why Traditional SOPs Are No Longer Enough

2:20 The Hidden Value Inside Your Existing SOPs

5:45 Why Most Knowledge Management Systems Fail

9:30 How AI Changes the Way Firms Access Information

13:15 Turning SOPs Into AI Prompt Libraries

17:40 Creating Internal AI Chatbots for Law Firms

22:05 Improving Employee Training and Onboarding

26:30 Organizing Firm Knowledge for Better Results

31:10 Prompt Engineering Best Practices

35:20 Common AI Implementation Mistakes

40:05 Balancing AI Automation and Human Judgment

45:15 Building an AI-Powered Knowledge Culture

50:10 Practical First Steps for Law Firms

55:00 Final Thoughts on the Future of Knowledge Management

Jason Marsh (00:01)
On today's show, Ernie Svenson, better known as Ernie the Attorney, and a longtime legal technology leader helping lawyers use tech to improve the way they practice. We'll be talking about how law firms can use AI to capture institutional knowledge using curated chatbots, and why replacing outdated SOPs with prompt-driven playbooks create more efficient workflows for training, delegation, intake.

and day to day firm operations.

Ernie Svenson (00:30)
Lawyers need to like calm down and focus on how to use this sensibly like all technology and then figure out how to get the most out of it because it can make your life a lot better. And if you're not using it 'cause you're you have this amorphous fear that something bad could happen, well, lots of lawyers are passing you up. You're in the slow lane.

Jason Marsh (00:48)
Welcome to the Alpha Lawyer podcast, a show dedicated to exploring what makes law firms thrive. I'm your host, Jason Marsh. Join me as I interview some of the smartest thought leaders and accomplished law firm owners in our space. Together, we'll dive deep into law firm operations, marketing, technology, finance, and everything else you need to build an incredible firm that allows you to live your best life.

Jason Marsh (01:15)
Hello, welcome back to the Alpha Lawyer Podcast. I'm Jason and joining me today, somebody I've wanted to have on the show for quite a while now, Ernie Svenson, better known to many as Ernie the Attorney. What's up, Ernie? Great to see you.

Ernie Svenson (01:29)
Yeah, it's great to see you too. I'm I'm excited to talk to you about all the stuff that I know we're gonna talk about.

Jason Marsh (01:35)
Yeah, no doubt, man. And shocker for everybody. We're gonna be talking some AI today. But Ernie, I've been reading through, you know, a lot of your newsletters that come out every week. I love reading them. And there's a theme that's sort of been coming up over and over, and that's the idea that most law firms, they have knowledge, they have some systems, there's a lot of expertise that they have within the firm to operate pretty efficiently. But the problem is

Is that the knowledge is trapped, right? It's buried in email threads, PDFs, SOPs that maybe nobody even reads anymore, shared drives, et cetera. worst of all, a lot of times it's just in people's heads, right? And it's it's just not anywhere to be found. So at the same time, we now have these AI tools that are becoming more capable and accessible than ever before. And the firms that are getting the biggest benefits are not necessarily the ones that are using the most, AI or different AI tools, but they're they're using AI.

capture knowledge, to improve their systems, train staff, and basically to eliminate these operational bottlenecks. So today I want to talk to you about what this actually looks like inside a law firm. We're going to get into AI powered knowledge systems, internal firm chatbots. That's a mouthful. And then and then one that I'm excited about, how to turn an SOP into a prompt library. But before we dive headlong into all that, I want to introduce you to everybody for people that may not already know you.

Which is a good size people that do. But Ernie, he spent nearly two decades as a commercial litigator in a large law firm for eventually launching his own practice. But for the past 20 years, Ernie's been l one of the leading voices helping lawyers use technology to build more efficient, profitable, and frankly, more enjoyable practices. Lots of recognition during that time. ABA Legal Rebel, Fastcase 50s, author of multiple books on legal technology. But today Ernie runs

The inner circle, it's his community where solo and small firm lawyers collaborate on technology, AI, automation, systems outsourcing, all things practice management. But what I love about it is that it focuses almost like an RD department for small firms, allowing lawyers, to test different things, share what's working for them, and just help everybody else avoid wasting time chasing every new technology, trend, or shiny object that comes along that we often do. And so Ernie, the thing I love

About you though, and I've known you now a long time. it's just that you've been on the cutting edge of legal tech for decades now, and you've seen multiple waves of technology and how it transformed the legal profession over this time from going paperless cloud computing, cloud software to automation. And of course, now there's AI. And so the first thing I want to get into is one of these ideas that you've written about.

and and we mentioned at the top there is that most law firms, it's not the AI that's the problem. It's the collective knowledge, the institutional knowledge within the firm. And since they already know how to do things and all this stuff, like we said, it's in emails, documents, SOPs are in somebody's head. And the thing is, is as a firm grows, that creates a bunch of friction you would expect, right? Team members interrupting each other with questions, new hires, takes them forever to get up to speed.

Partners often are bottlenecks 'cause they're only the ones that know how to do a certain thing. And so f Ernie, when you look at solo and smart firms today, like how big of a problem is this hidden knowledge problem, if you will, and how have you seen that manifest itself?

Ernie Svenson (05:00)
Yeah, well I think it's not even just a problem with solo and small firms. It's anywhere people are collaborating and doing knowledge work together, you have the potential for them not to be able to share their knowledge efficiently. I think the biggest disruptor was I mean paperless is kind of what made it possible, but now we all work remotely, yeah or for the most part, there's a lot of remote work, even in big firms.

COVID was an accelerant for that, but that was kind of already happening anyway. we all know that story, right? But when you have people that are working together but they're not in the same location and they can't and they don't meet every day and talk about things where, the the boss can bellow down the hall and the and the secretary of paralegal comes running, that

way of operating, I think, hid a lot of the lack of knowledge sharing, right? Because you can rapidly make up for it by just calling somebody in. But when you can't do that, because everybody's distributed, you really have to think about how to share this information asynchronously as well as remotely, then y you start to see the problems. Like the problems become manifest quickly. So that was that's where it started and it still exists because humans are

Jason Marsh (06:10)
Well th it's it seems

like this is a problem, even if everybody literally worked in the same office and obviously many many do, this problem still exists, right? 'Cause this guy knows how to do that and somebody else knows how to do another thing.

Ernie Svenson (06:14)
yeah, the yeah, yeah, yeah. Yeah, no, no, for sure. No, no, for su for sure. But

it it's easier to pretend it's not as bad or it's easier like I said, if you if you're lawyers are notorious and have always been for doing things at the last minute. So if you can cover your ass by, pulling somebody at the last minute, no harm, no foul.

But there is a harm and there is a foul when you can't do that. And so I think that's exposing the problem. And the other thing is once you can start automating things, you can only automate that which has been systematically worked out, right? You have to say, okay, if this happens, then we're gonna do this, then we're gonna do that. So that systems thinking, the rigorousness of that

gets exposed also when you're automating. And l a lot of lawyers like to think they're very rigorous thinkers, but in reality, if you tell them, okay, well if you're such a rigorous thinker, show me the flow chart of the decision tree and everything else. yeah. And then also these vagaries come out and they think that they're saying something perfectly, clear and methodical, but in reality it's not. So what lawyers and not all of them obviously, but what many lawyers think is true because it's true in their head

but reality exposes that it's not really true out there. that's the problem. And technology has just made that increasingly obvious that people are not as rigorous in their thinking as many of them like to believe they are.

Jason Marsh (07:35)
Yeah, and it's funny because we were talking about this before the show for a minute, but the idea that, AI and technology is gonna solve all these problems, but the reality is is that, automation, technology, AI, these are just accelerants. So if you have, jacked up systems or none at all, or other, not rigid processes for doing things, all that AI and tech does is it just makes all those problems worse. I've heard this talked about.

in other places too is solve those things and then use AI to scale all of those things. And so but let me ask you this, like before we get into some of the more detailed, things that we want to talk about, but like when a firm does this well or solves this well, what does that look like for them all of a sudden?

Ernie Svenson (08:18)
well, I mean it l it looks like Nirvana compared to the opposite, right? I mean, it it's I don't know, I guess I've told this story many times, in my podcasts and other places. Like I was very fortunate when I got out of law school. I spent two years clerking for this federal judge and he was a federal trial judge and he was very systematic. Like he had a book called The Bible and that was a book that was designed to tell you how to do everything. I mean and the detail in it was shockingly,

it was shockingly detailed, to the point, for example, that there was a section in there about which of the two clerks would water the plants and when. Now, watering plants is not an incredibly important judicial function, but we had plants and they would die if somebody didn't water them and he wasn't gonna water them. And that was the whole premise was like, I'm not doing all this stuff. We're paying you. You guys do this. So there was nothing that was part of my job

that wasn't in the Bible. And if something came up, because inevitably, work develops in some way and there's something that should have been in the Bible, or maybe it should be in the Bible, but it wasn't in there before, he'd say, Well, go, go check the Bible. Say, Well, Judge, it's not in there. He goes, Well then put it in there. So he didn't even update it. And so that what it looks like if it's done right is person at the top is doing very little other than orchestrating and everybody below knows exactly what they're supposed to do and when.

And it gets done and occasionally things, fall between the cracks and you address that. But for the most part, things work really well without, him telling you what to do. Like the first day I showed up at work, he didn't give me like a hey, welcome, here's what you're gonna do. It was like there's a book on your desk called the Bible. Read it, you're responsible for everything in it. And I thought everybody operated that way. So that so when I went into the practice of law, I was expecting to see something like that, although I kind of had a sense that that probably wasn't the case.

based on a lot of the lawyers who appeared in front of him. But I was just like thinking, this is what y sophisticated firms do, and and good lawyers do. And the shocking realization was, no, most of them don't actually do that.

Jason Marsh (10:18)
Yeah, it's funny you mentioned Nirvana as you're describing that. And I think any whether you're a a law firm leader owner or just an entrepreneur, business owner, what you described is, ⁓ there's a book and it describes how everything needs to happen. You don't need to touch it, they update it and so yeah, I love that. So anyway, your answer almost as if like you knew what my next question was gonna be, but I do want to talk about, one of the solutions to getting to that point that you've been talking about.

And then it's the idea of using AI to make the firm's, knowledge accessible to everyone. And so you've written about, and I think you've even created your own within your program, these curated chatbots, these systems, what is it, RAG or RAG systems, and using tools like Notebook LM to create what's essentially an internal AI powered help desk or a version of the Bible.

Ernie Svenson (10:51)
Mm-hmm.

Jason Marsh (11:09)
so for someone hearing all these terms for the first time, what exactly is a cured chat bot and why is this really such a big opportunity for law firms that don't or even the ones that do have the Bible?

Ernie Svenson (11:22)
Yeah. Well the th the thing about creating so what what people need is an SOP, like a standard operating procedure. there's a famous book, Checklist Manifesto, by Atul Gawande, who was a physician. He talked about how a lot of the malpractice or mistakes that happened in medicine was because somebody missed something on the checklist. Pilots, famously have checklists for everything. I took flying lessons at one point and it was shocking how like they'd say, Go check the ailerons.

go do all this stuff and you think, really, do I need to do this every time? And the answer is yes. If you want to be methodical, you need to do certain things the same way every time, right? So creating the thing about creating SOPs is it's boring, it's drudgery, it's it's annoying, nobody likes doing it. And then maintaining it, same thing. But if you have them and then you expect people to use them and they do, that nirvana that we talked about happens earlier. How AI can help that.

unfold better is that it makes it less annoying. Like if you go to AI, a chat bot and say, okay, I need to create an SOP for intake. I'm an estate planning lawyer. Here are the kind and you just have a brain-up of all the things you think of. They don't have to be organized. This is the beauty of AI. You can take rambling, unstructured off the top of your head mumbo jumbo, which equivalent of spaghetti on the wall, and it'll organize it and put it into something

Kind of structure. That's what it's really good at. that's its highest and best use, most reliable use. So you can use it to create the SOPs, you can use it to modify the SOPs. And at some point, if you've got this all written out, you can give it to the chatbot and say, okay, your job is when we do this to walk us through it so that it's done properly. A chatbot walking you through a set of instructions that it's developed with you.

is going to be more fluid and dynamic and adaptable than a rigid recipe. So that's the first big benefit because it can spot things or think of new things on the fly and it can and that's just, that's very desirable. So it makes it more enjoyable, makes it work better, and you can, create the SOPs faster. So

Jason Marsh (13:32)
And

you can dialogue with it, right? Whereas obviously a book or a document, it says what it says. You can't ask it questions, or at least you couldn't. Now it's like, okay, I've read this. I don't fully understand what does that mean or what is this acronym or what is it like I mean you that's the part I love just about using AI and learning anything in general. I'm not just scrolling through a list of links on Google trying to find the article that roughly addresses what I want, but I can literally just keep asking follow-up questions more and and get deeper and deeper.

Ernie Svenson (13:34)
Correct.

Yeah.

Yeah? Yeah.

Right.

Jason Marsh (14:00)
So to to this extent, now when you say a chat bot, is that what Notebook L is, right? There's a Google product. Is that essentially the tool you recommend in creating this thing? Or just help us consolidate, you know, this bigger idea into practicality.

Ernie Svenson (14:11)
Yeah, yeah, I mean yeah.

So AI, there's two kinds of AI. There's deterministic AI and there's generative AI. What everybody's talking about these days is generative AI, the chatbots, the things that can generate thoughts and have a dialogue and so forth. we'll forget about deterministic, but deterministic is very mechanical and more rigid and it's gonna produce the same result every time. So it's basically like automation. But this generative AI with a chatbot, yeah, that's Gemini, which is Notebook LM, Google's tool.

the there's chat GPT, there's Claude. Perplexity has its own kind of minor one, but then it also calls upon the other major ones. So it doesn't really matter for our purposes as we're talking now to identify one over the other. It's just you're having a conversation and working things out and that's that's beneficial. There's there is one other thing too to mention, which is when these things first arrived they had a cutoff date and they couldn't search the internet.

Now, even though they have a cutoff date for their training, they can go find the answer on the internet. And this also makes them incredibly valuable because if you're asking it a question about something and it can go find the answer on the internet, that's really good. Like I mean, I used to use Google to answer all kinds of problems about troubleshooting technology. And that was very cumbersome to pull those answers out because first you had to get the best answer somewhere, then you had to read it, read through the ads, it's a whole thing. Now you just go, hey, my computer won't do this.

tell me what to do and it'll tell you. And then if you realize like you're missing something, you just tell, Well, I need to, give you this additional piece of information. So it it's the power of being able to look things up on the internet and converse with you as you think of more things to tell it or additional things that need to be told, this is just huge. that if it did nothing else than that, this is amazing. But there's more that it can do. But that alone is

the table stakes and everybody should be using it as much as possible for everything. Obviously it can make mistakes, obviously it doesn't it's not perfect, but neither are humans, by the way. So just keep your thinking cap on your head, sensibly skeptical, and you're fine.

Jason Marsh (16:26)
And so yeah, I it's funny, I saw an email you'd put out this morning just talking about, lawyers, the headlines of lawyers getting in trouble for this or that, but it's not the reality and it's no different than it was thirty years ago, is if you were taking, citations or some sort of input from an associate or an admin and didn't bother to double check it yourself. Like other people can get things wrong too, and at the end of the day, you're an attorney, you're responsible for what you're submitting to whoever.

clients, courts, and so forth. so if you haven't reviewed it and confirmed the facts, that's on you. But in terms of these curated chatbots and building these AI knowledge systems, is the idea that this is a repository. Just dump everything you have, whether it's around a particular topic, and I don't know exactly how you can structure these, but you put it all in there and now

it's in your AI tool and you can query it and it's gonna look in these documents and maybe videos, all the different sources that you give it. I know I do something sort of like that within Chat GPT. You create a project, you add all the sources in there, and then you can create different threads around different things, but is this roughly what we're talking about? in building, this firm knowledge system?

Ernie Svenson (17:33)
Yeah, so it can go get information on the web. We know that. It's got its training, that's great. But yes, you can give it information and say, I want you to help me examine this, consider it, look at these different things. So you can focus it in on a repository. And the the acronym for that is ⁓ RAG, which is stands for Retrieval Augmented Generation, which sounds completely vague and n doesn't make any sense, but

It is the it is the term, right? So you focus it on something. So Google's notebook LM is like the best example of this because it you can't do anything with it that it isn't a rag. It's all a rag. Like you put the stuff in, it doesn't go look anywhere else, it focuses only on what you give it. That doesn't eliminate that it would hallucinate, but it almost completely does. But what it still can do is maybe misunderstand or not accurately represent something.

So you have to check its work as well.

Jason Marsh (18:28)
Can you do that? Can

can you ask it to cite where it's giving you an answer? Like point to the document so that I can go in that document and read it.

Ernie Svenson (18:32)
no, it does. It no it does.

That's what yeah, yeah, that's what it does. You put all the stuff in there and then when it comes up with an answer, it's all those answers are cited. So it'll you you can hover over the the source, click on it, it'll pull up the source. So you can instantly verify what is being said and you know, see for yourself it if it was accurately stated or represented. So Google's notebook LM is I think for lawyers definitely a tool

that should all they should all investigate it. And it's it's like I think the least expensive of all of them. Most of them are like twenty dollars a month. You definitely don't want to be using the free ones. And I think Google to use notebook LM is like fourteen a month maybe. So it's a little less than the rest of them. But I I definitely recommend that Lauriers look at Google Notebook LM and understand that one because it is nothing more than a rag. You can make the other ones, focus in on rags, but you're never quite sure unless you're good at this,

what it's looking at or not. So Notebook LM is super powerful. And and Google's AI engine, which is Gemini, is has long been held to be the best in terms of general capacity and capability. So it doesn't mean it doesn't hallucinate, doesn't mean, the other ones aren't as good. It's just Google knows how to do AI. And of course they do. So check out Notebook LM for sure.

Jason Marsh (19:49)
And and so I think probably you and I both know the question as you say this, you're dumping all sorts of information and what's popping up in lawyers listening is confidentiality, are there concerns around that? Is that data protected? Is it training the AI like is like their private data gonna show up in somebody else's answer somewhere?

Ernie Svenson (20:09)
Yeah, I mean the explanation I always give is this, and is this a good way to conceptualize it. If you think of what these things have been trained on, they've been like the amount of data, let's just pretend it's like the Pacific Ocean, okay? Like that's how much data it is, right? And then the additional information that they could get from the web is more data, right? So all of that data and then you take your da what's your data in comparison to that?

It's like a thimbleful or an eyedropper drop. That's what it is. So if you think of how likely would that be like, bottle message in a bottle in the ocean to pop up on an island where somebody discovers it, it's like it's not gonna happen. That's not how the problem is gonna occur. The way the problem could occur is if something very specific, like that's of high value, like some topic that people are interested in.

And you have something very specific. That very specific piece of information could surface as the, it as part of a response. And maybe somebody could figure out, well, you're the only person who had it. But that's a lot of ifs, right? That this is we're talking like Carl Sagan a astronomy, like level of unlikelihood. So that's not really the risk. The risk, and here's where I wanna remind lawyers, like while you're busy looking off in this direction where it's highly unlikely to happen.

You're taking your eyes off the most obvious place where it's gonna happen, which is you or the people around you misusing it by making their passwords too easy to guess and their data is stolen and the usual stuff that happened with the cloud. And even for that, lawyers have not been even as reprimanded as far as I know. Like it it's not good, you might have to disclose it that you were an idiot and l let your data out. But nobody's been even reprimanded. So

Lawyers need to like calm down and focus on how to use this sensibly like all technology and then figure out how to get the most out of it because it can make your life a lot better. And if you if you if you're not using it 'cause you're you have this amorphous fear that something bad could happen, well, lots of lawyers are passing you up. You're in the slow lane.

So put your foot on the gas and learn how to drive carefully and you'll be fine.

Jason Marsh (22:18)
Yeah, it's a that's a good one. I like that idea. Yeah, just you gotta use it, but just be careful about it. so just quickly, how is this idea of a curated chat bot different than just simply using Chat GPT? Is it because it's a rag system you're putting your own information in there? So it's using both the wider training knowledge base as well as your stuff?

Ernie Svenson (22:39)
Yeah, that's basically it. You it's it's operating inside of a very tight boundary that you've given it if you're using Notebook LM or focusing on a rag, as opposed to just asking it questions. Yeah, that's exactly it.

Jason Marsh (22:51)
okay, cool. All right. So we got that. Now let's talk about what are some of the practical examples you recommend and how a law firm might use this, whether it's, a new higher onboarding, you mentioned just creating an intake, process, SOP, FAQs, workflows, just whatever general policies around the firm. What what are some of the best use cases that you like to talk to people about?

Ernie Svenson (23:12)
Well, these days and this is we've gone from in my inner circle, like from a year and a half ago, we were talking about Chat GPT mostly 'cause that was the dominant one. And yes, Perplexity and Notebook L and Claude, we kind of like, yeah, that's like ChatGPT, but if you prefer Claude, use Claude. And Claude was always slightly better at writing. But more recently, Claude has gone to the fore because of Claude Code, Claude Cowork, these various tools.

And it's widely acknowledged or there's a lot of consensus around the idea that Claude is better for most things than Chat GPT. Now, there's some I'm not saying for everything, but f for the purposes of lawyers, Claude is better. Like there's there's a f a famous guy named Zach Fowler, I think is his name, who talked about the Claude native firm and people went crazy about this and it was a LinkedIn post and

He was basically saying, I use Claude to like run my whole firm. He's a solo. Now he's that's somewhat exaggerated, but he's he's using it a lot. so Claude is the first thing I would say lawyers need to zero in on. What can Claude do? Well, we probably don't have enough time to cover it all, but the number one thing that lawyers should focus in on with with Claude is something called skills. What are skills? Skills are let's say you're doing something.

and you're having a chat back and forth about onboarding or, answering discovery, something that's a routine task, a normal task, and you're talking through how you want it to do what you want it to do. If at the end of this chat you say, Hey, listen, Claude, I think this maybe is a good skill, you tell me, 'cause and I and I tell people this all the time, just treat it like you don't really know that much and let it guide you because it's really good about not steering you

in a bad way. Again, keep your thinking cap on and all that good stuff. But if you just say, hey, do I do I need a skill for this? Yeah, let's build the skill together. And then it'll then ask you questions like, well, do you want it to be like this? what it'll ask you things you probably should have thought of to tell but to tell it, but you didn't. And it'll draw all that out. It'll get a really organized sense of what this kind of task that you're giving it want it to create a skill for. And so basically a skill is like an SOP. Again, we're back to the SOPs.

But it's a particular kind of SOP that's really neat. Because once the skill is built, Claude will say, okay, great. Now go ahead and hit save and you'll save it. And now it's stored up in the servers in Claude. And when you ask it to do that kind of task again, you won't have to say, hey, go get that skill. It'll know from the context of what you're asking it to do. it'll always be looking. Like, do I have a skill for this? look, I've got a skill for this. And you can see it saying, ⁓ accessing skill, thinking.

Looking at this, looking at that. So the skill now becomes the thing that moves the ball forward automatically and in a better way. And as over time you realize like, well, maybe that skill didn't work exactly like I wanted to, you tell, hey, Claude, listen, like let's update that skill. Here's some things to think of. It'll interview you, you'll go okay, fine, save the new skill. So it's even easier to create these SOPs if they're in the form of a skill. But then it gets even better because if you're using the team's version of Claude, where

Multiple people can all be working off the same account where they give you some additional protections. That skill can become an organizational skill and is now available to everyone on the team. And that is what leads you into the place where now you're using this tool in a much more sophisticated, beneficial way.

Jason Marsh (26:45)
And so y what would be s just a couple of ideas for for anyone listening. What are some some good skills, like, you know, what good first project or skill? What are some good uses for this so people can conceptualize what you're describing?

Ernie Svenson (26:52)
Yeah.

Yeah.

Yeah, and this is the question that comes up all the time in the inner circle. I'm well prepared to answer this question. well first of all, conceptually, it's anything you're doing repeatedly. Well, what is that? Writing, obviously. You're always writing. And so I tell people, you should build a skill that can write like you. Now, you can get really granular with this and tell it, okay, I want you to write I want a skill for when I write emails, and I want a skill for when I write a brief.

that's not it's not a good idea to try to specialize with the skill at first because it's your first skill. So it's better if it's general. It's better if you maximize its its use in a general way. And over time you may say, it's fine the way it is. I can use it for writing anything because there's a general way that you write. That's certainly true for me. Like there's a general way that I write most things. In my case over time I realize like, well, but when I write a Substack post

I really want it to be considering this. So I have a separate skill for the substack post writing. And then same with LinkedIn. LinkedIn is a different, it's mostly the formatting, but there is a an approach to it. So I trained it on LinkedIn. So when I write something, I'll say this is a post for the inner circle. All right, let's turn it into a substack post, it knows what to do. I'll say, okay, turn that into a LinkedIn post, it knows what to do. And none of them are so

Good that I look at them and go, that's exactly how I would have written it. I'm not changing a damn thing. But it didn't take me long to get that first part. I'm just reacting to things and changing things based on a reaction. That's not heavy work. That's not hard to do. So it makes it really easy and enjoyable to write these different things. So the writing skill is number one, I'd say an email parsing skill, like one that where if you connect it, okay, this is another concept. If you take these tools and say, I'm gonna connect you to my email.

Jason Marsh (28:18)
Right.

Ernie Svenson (28:45)
Whether it's Outlook or Gmail or whatever, and say, I want you to look at my email today and tell me what I've got that's important. It'll look at it, it'll break it down for you. And y you might say, as I did before I created the skill, Well, why don't I just go look at my email? Well, the answer is because there's a lot of noise in there, and you have to cognitively sift through the noise. There's something nice about having it handed to you on a platter by a very helpful AI tool. And then and again,

Jason Marsh (29:14)
It's almost like you have

an assistant. It's like reading your email and surfacing the most important things. This is what you need to, focus on. We can potentially handle the others. I even was saying early on I was, AI, I was just calling it an assistant intelligence. I was just treating it like a personal assistant and trying to get a lot of low value stuff or time consuming things that it could do reasonably well, to do it. How difficult in terms of setting up these skills and Claude or something similar

Ernie Svenson (29:16)
Y it is like exactly yes. Yes.

Jason Marsh (29:40)
and is it similar is like a custom GPT somewhat similar to a skill in Claude versus

Ernie Svenson (29:44)
It

it it is. The the difference between a custom GPT and a skill is the custom GPTs don't fire automatically. ⁓ that's the biggest one. but otherwise it's the same concept. Like it's a s it's an instruction set you're giving it that it stores. And what's interesting is you can and I remember early on when skills came out, people would say, Here's how you build a skill and I'd look at that and think, Man, that seems like a lot of work. I'm not doing that. And then I learned, well, you don't have to. You tell it what you want, it builds the skill, it stores the skill.

And you can look at what it stored and say, okay, it captured it and you can examine it. When you examine it, you'll see there's a structure to it, and it's basically just like the custom GPTs, which you also can create by asking it questions. Like it it stores the instruction set automatically there. You can edit it, but for the most part, it does all the heavy lifting. So yeah, pretty much the same thing.

Jason Marsh (30:35)
and does it work well? Cause I have thought to myself when it comes I've created loads of custom GPTs with very detailed instructions within them. And I'm like, if this custom GPT was a human, I would have fired them twenty five times over because they're constantly screwing this shit up and it drives me crazy. I'm like, the instructions are there. I spent the time creating them. Why are you not following it? I said three bullet points, not ten or whatever. I so I haven't used skills in Claude, but

This sounds encouraging. ⁓

Ernie Svenson (31:05)
I think

I think skills in Claude are better at least than what I remember custom GPTs to be 'cause I haven't created one or used one in a long time. But they all can fail to follow an instruction. and you just reprimand it. But I think they're all getting better at that because the developers know that they have to make that better. So they're all getting better but they're not again they're not

Jason Marsh (31:14)
So

Fair enough. All right. So I want to get into the next thing that I'm really interested to talk to you about. and this is an idea you've written about. and like I said, it it caught my attention. I think it's a really cool concept. We've talked about the importance of SOPs. Every business owner knows how important they are, but you have an idea of replacing sort of the traditional SOP or Bible that the judge you worked for had with this idea of prompt libraries, right? So I think

Most firms have some version, like you said, operations manual or document. but the thing is these things get outdated or nobody's even reading them or nobody owns them to keep updating and maybe sometimes they're just not even capturing how work actually gets done anymore. And so that, what you're saying is AI can fundamentally change how firms document execute these processes. And so what do you mean exactly when you say prompt libraries could become the new

SOPs or operations manuals.

Ernie Svenson (32:21)
Well, first of all, let me say I have to update that statement because I think the new SOPs are skills and prompt libraries are good, but they're cumbersome and you have to store them, you have to manage them like a librarian, and they're not as easily shareable and they don't execute automatically like skills do. So my mantra these days is skills are the way, the truth and the light,

And when you combine skills with a gentic AI, where it starts doing things on its own and works with other agents, you gotta have something more like skills to make that happen. So a prompt library is a good thing to think of as a repository of your best prompts for different things. But where that should live to me now is in skills.

Jason Marsh (32:45)
So

And and essentially a skill is just one big set of instructions, just a big long prompt of a repeatable task. Then as you mentioned, it can be available either just to you or it can be available firm wide or company wide. so either way, conceptually I think it's the same thing. Tactically, it's a little bit different from a prompt to you know a skill, maybe even just semantics, but

Ernie Svenson (33:13)
Yeah. Yeah.

Jason Marsh (33:28)
But how do you go from, all right, I've got all these SOPs documented and wherever they are? How do we move that and convert that into prompts and skills and these automated ways of doing that? Is it just dumping them all into the okay?

Ernie Svenson (33:42)
No. No.

Dumping bad. It's like ⁓ reminds me of that stupid s it reminds me of that Saturday Live thing of Fire Bad. it was it no ⁓ Tar the Tar anyway, the the Tarzan thing, you know, Joe Piscava was it? Fire bad. It was Tarzan guy. Anyway, ⁓ the the dumping things is bad. And this is

Jason Marsh (33:47)
No dumping.

Dick in a box.

Ernie Svenson (34:06)
This is a phenomenon that happens constantly and it comes from the basic human nature of like, I just wanna get this over with as quickly as possible. I'm gonna dump all that stuff in there and let that tool figure it out. in theory it can and it can figure a lot of things out, but it's a bad way to operate generally because of the principle of garbage in, garbage out. Like and it's not gonna throw as much garbage out as it would if, in the previous days where it was, just automation mechanistic. But

It's just not a good way to operate. So, but it's a it's a persistent thing. And I see this with the lawyers I work with all the time. I'm like, look, you're trying to do too much too fast. And I know because I do it myself. But now I catch myself, probably more so because I'm telling people not to do it. So I hear myself, telling other people do it not to do it. Now you're realizing, you're about to go down this trail. Don't do that. So it's and this is basic engineering. Like if any engineers are out there, software developers, and they go, Let's make this

So that it works for a hundred thousand people. They don't go, great, let's just release it now and have a hundred thousand people use it. They go, let's test it with one person, now with three. we found some bugs, let's fix them. let's try it with ten. so a few more bugs, let's fix that. You iterate very slowly and methodically. That's how engineers operate. And here again, I think lawyers think that they're very methodical like engineers, but sorry folks, you're not. to watch true engineers

They don't operate that way. And I know, Facebook's mantra was move fast and break things, but they're moving fast in a methodical way. And what they when they say break things, they mean don't be afraid to let things fail so you can see what failed and fix it. That's not what we're talking about. We're talking about being very careful. And if you do that, a couple of things happen. One, you don't stress yourself out. Two, you get to adjust those things, with more time and thoughtfulness.

And so the results are always better all the way across the board, right? The problems come from trying to do too much too fast. So don't do that.

Jason Marsh (36:00)
Yeah, and one of the things I think you've said that I liked is the idea that focus on outcomes rather than the specific processes

Ernie Svenson (36:08)
Yeah, yeah. Well, I mean the outcomes tell you if it's working or not. I mean, you should be process oriented. But i the nice thing about AI is if you don't like the result it gives you, tell it some tell it to do something different. there's somebody I recommend that people follow named Allie Miller. She's on LinkedIn, she's, a guru of AI. Definitely follow her. She's got a lot of practical tips. But the one that she issues forth repeatedly is straight from the Spice Girls.

that song, tell it what you want, tell me what you want, what you really, really want. That's what you gotta do. And that's where people are failing. They'll say, it didn't do that thing. Well did you tell it? Did you tell it you didn't like the result? You know? Did you continue the conversation? you didn't. Okay, well, how is it gonna improve then?

Jason Marsh (36:51)
Yeah. And so let's talk about s building out some of these skills or, prompts thing conceptually. What kind of things can they be used for? Like, intake, client onboarding, training new employees. Like talk to me about some of the the cool skills that you see some people in your program using and having a lot of positive experiences with, both in terms of actually working the way they've intended it, but but getting them, you know, some significant time savings.

Ernie Svenson (37:02)
Yeah.

Jason Marsh (37:18)
accelerating what they're able to do in a given day.

Ernie Svenson (37:22)
Yeah. So the things that AI is really good at where you have less fear about it doing something, catastrophic, is like summarization. Like so the note taking tools, which and I wouldn't try to create your own, like they they're out there, like like granola's one I like and use it, but there's multiples. Let it take notes, right? Like that's a wonderful thing. Like let it take notes because it's capturing information that you might want to use.

Then when you need to go use that information, you can either take the summary that's been generated or get the transcript and feed that into a new tool and ask it to do things with it. So monitoring your intake of of knowledge in the form of conversations you're having with people or meetings you're having, that's low stakes, but high value. I mean, low risk, high value. from there, we get it, and then that's where onboarding would come in. So, like let's say

you're onboarding new clients and you're saying, well, what happens in onboarding? Well, they tell you what their problems are. Well, you have to convince them that you're the right lawyer. We have to tell whatever. There's a bunch of things that happen in an onboarding introductory meeting. So one, you should be recording those. Two, tell the the prospective client you are recording them. Make sure you preserve the confidentiality and don't do things that that don't share them with non lawyers that are not, you know, privileged. Do the obvious things, but then

If you take these sessions that are occurring across multiple clients and you ask the AI, hey, listen, here's like ten sessions. What do you see as the patterns? What are we doing well? What can we improve? That's not necessarily even something you need a skill for, although you can build one and say, Here's what I want to know, but it's gonna give you insights that are incredible. And so there's just a whole bunch of stuff it can do. It can it can help you prepare

For direct examination or cross examination. Again, feed it all the stuff for the witnesses and say, Hey, I need to set up my questions. Can you do that for me? It'll do it. Now, what's gonna happen is you're you're gonna see some of the questions and think of more questions, great. You're not gonna like some of the questions, fine, change them. But the amount of BTUs, cognitive BTUs that were expended in creating this are like nothing, right? So you can

very quickly get to a place where you're able to do the next important thing if you do those would be good examples right there.

Jason Marsh (39:41)
So another theme that comes up a lot in your writing, which I really like, is the idea about AI the wrong way and in and where people treat it like a search engine or a vending machine, which is funny, or just a general piece of software or something. But what you're saying is that a better model is don't think of it as a tool, but think of it as an

Associate, kind of like I was saying before. I was just thinking of this thing as an assistant, just do these basic tasks. But it seems like a simple idea, but it has a ton of implications, right? For how lawyers can communicate and the results they can get from AI.

Ernie Svenson (40:06)
Mm-hmm. Mm-hmm

Jason Marsh (40:17)
Ernie, what are the similarities between, would you say, managing an associate and then managing your AI tool to act like an associate?

Ernie Svenson (40:26)
Mm, good question. well, I think it's giving it direction, giving it good direction, but back to what I said earlier, telling it when it d when it didn't give you what you wanted and being very explicit about exactly what it was that you wanted and that you didn't get, right? The more detail you give it about what you expect it to do or what you the result looks like that you want, the better it can get at

doing it. One thing that does happen like so for example with writing, which is that's what I do mostly and that's what I care about mostly. So I never thought of trying to describe what I do when I write or what I care about or don't. But as it gives me things, I'll say, I don't like it when you use that word. Like that word annoys me. I don't use that word. And that's a that's focusing on a word, but then it'll have patterns.

And there's the famous one like it's not X, it's Y. But there are patterns that it will use or that I will use and I'll say, I didn't realize that I used that pattern. So you become more aware of the patterns you're using in certain things as you look at them in terms of skills or whatever that it's creating. So just pay attention. I mean the key is to pay close attention to what's coming out.

And I think many of us would rather not. We just like rather move on to the next thing. But that's what it takes, right? Like you give you give feedback. It's a feedback loop.

Jason Marsh (41:48)
Yeah. Well it's it it's

it's yeah, feedback and context, right? You've talked that context is really important and the idea that you can create context documents. How how does that work?

Ernie Svenson (41:59)
well, the first thing that comes to mind when you say that because this is where I find it most helpful now, is there's this Claude has this tool called CoWork, which is a desktop app. Mac PC doesn't matter, it's on both. And with co work, you can aim it at folders. So you can say, I'm gonna have these folders, and I know I'm gonna want it to work with these folders, and you can switch gears and go, okay, look at this folder, and then let's talk about.

things we need to do. Well in the folders is the context, right? The folder is a rag, but you can also put into the folders little instructions. They're markdown files, so they end in the extension.md. And generally the top level one of these instruction sets will be called Claude.md. And in that folder or in that dot that in that text document will be the instructions and things you want it to do for that particular folder.

So it's kind of like a skill, but it's different in that the skill is living wherever it lives and it gets called upon automatically. In this case, it only gets used if you're aiming it at that folder. So those little instructions are one thing. The other thing that you can put in those folders is a memory file, and that's what that would be memory.md. What what's in the memory the in the memory file? It is a memory of what important things were done.

through the sessions. So instead of it having to look across all the chats and read all that stuff and use up a bunch of tokens and take up a bunch of time to figure it out, it creates its own little trail of breadcrumbs, memory. And that when it goes in, it goes, okay, I know what we did. I know, that whole history in terms of that context, plus the context of the documents. And your job simply is to orchestrate that by letting it create these little files and then, managing them.

Jason Marsh (43:48)
Yeah, you can see how doing that would make a massive amount of difference because you're giving better input on what you actually want. And again, better input, better

Good I don't know what the converse saying is that. Good in, good out, sounding like that, let's talk about the future, Ernie. Do you want to talk about the future? ⁓ you know, in terms of writing a law firm. But obviously we we've talked a lot in this conversation, right, about what lawyers can do today. But again, like the thing I love is you're you've been on the edge of technology and legal tech specifically for decades now. As we said, you were

Ernie Svenson (44:06)
Hmm. Sure, why not? You wanna talk about aliens?

Jason Marsh (44:25)
one of the first people talking about moving to paperless and then of course cloud computing, AI automation, AI now. But what's look five years? And and I think that the other thing too, Ernie, is while you've been doing this for decades, like these decades have seen the most insane amount of technological advancement in human history. And it's only compounding right as technology keeps evolving. It happens more faster, etc. But five years, I mean I and honestly I feel like

Ernie Svenson (44:50)
Yeah.

Jason Marsh (44:55)
it is a millennium away in terms of, things happening. But at the same time, law firms, lawyers tend to adopt a lot of technology in a slower, let's say careful or methodic way. So when you look like five years into the future and you think about, and again, this goes back to what you're doing with your people in your community, what will be happening inside the best law firms? What what will that look like?

Ernie Svenson (44:57)
Yeah.

Mm.

Hmm. Well, if AI is gonna keep advancing at a rapid pace, it's gonna depend on energy sources. So there's some assumptions there that are foundational. But assuming that everything keeps, humming along and evolving, I think what lawyers do is pattern recognition and critical judgment, right? So the pattern recognition part can be handled by AI. The other thing with

When you're doing pattern recognition, the next step after some pattern is recognized is do you act upon it? This is where agentic AI is kicking in, because agentic AI says, there's a trigger, or I sense something happen, which is the trigger. All right, now I'm gonna do this because of that ha thing happening. And then it can branch off and it can do this in very sophisticated, complex, amazing ways, right? And so all of that heavy lifting of the pattern recognition, the investigation, the research.

the the fine tuning of the research and the and the ch f site checking the research, all that can be done by AI. What humans are gonna be needed for is the judgment. The like, you know, given all of those circumstances, I think what we need to do is this. And here's why we need to do it. And making the humans you're advising feel comfortable because they're not gonna wanna just I mean the sophisticated humans are not gonna just wanna do whatever an AI bot tells them to do for

For reasons that are not discernible to them. So that people are gonna want to know that they're getting good advice from somebody with good judgment. They're also gonna wanna know that that person is not wasting time and energy gathering all that stuff themselves and nor wasting money. So lawyers are gonna have to change their profit model probably to s a large extent. I don't know if the billable hour is gonna die instantaneously or quickly, but.

know, the only reason why we had that was that was the best way to estimate or to price the work that was being done. the question is really what's the value? And people will pay for value. That's what they want to pay for. So lawyers are gonna have to adopt ⁓ a new way of operating and some will and some won't, and the ones that don't are gonna fall behind and eventually their way of doing things won't be done. That's what's gonna happen.

Jason Marsh (47:34)
Yeah, it's interesting because what you're describing here, the promise of AI over whatever, five years, twenty years, is the same promise as technology going all the way back to like the invention of the wheel. It's just there's a certain part that is gonna become a lot more efficient, but then there's human judgment and guidance that's still, well, hopefully, will remain important. But if you had to I don't know, pick out a s a specific role within a firm.

Ernie Svenson (47:46)
Mm. Right.

Yeah.

Jason Marsh (48:03)
that's gonna become more valuable in these type of AI enabled firms. What would you say that would be? And when I say role, it doesn't even have to be like a job title, but a specific role that someone may play as just part of their job.

Ernie Svenson (48:09)
Mm-hmm.

well it's gonna be what the IT head was the the equivalent of that. It's gonna be a person within the firm who knows how to set up and manage the agentic workflows because that's what's gonna be driving a lot of work. And so those are gonna have to be done well. The agentic stuff is gonna have to be coordinated and supervised in a certain way and updated and all of that. And lawyers with traditional lawyer training are not

ready for that. Even IT people aren't really ready for that. So the person doing that job is going to be somebody who's up to their neck and AI, using it all the time, and getting a sense of how it works and staying on top of that. They will have earned the trust of the people that hire to do it because if they don't have that trust, they're not going to be given rain to do what they need to do. but that's going to be the critical job role. And there and there'll probably be more than one person doing it, but there's going to be at least one.

And and maybe there won't be more than one. Maybe one's all you need because most of the work's gonna be done by the AI stuff. But that's gonna be the essential ingredient in any kind of knowledge work firm, not just law, any anything where knowledge work is being done, that's gonna be critical.

Jason Marsh (49:30)
Basically it sounds just like a systems administrator, like someone who understands how all these systems work and how to implement them in whatever particular business and set of use cases are are existing out there. So listen, we've we've covered a lot today from like these knowledge management systems, AI powered chatbots and skills and prompts and

Ernie Svenson (49:33)
Yep. Yep. Yep.

Jason Marsh (49:50)
delegation of all that stuff and and the future of running law firm. But just just for a law firm owner, leader, attorney listening to this right now who's like, I want to take some action. what should be the next thing? maybe they dabble or like you said, they're treating AI like a search engine or a vending machine. But what's the next step they can take?

Ernie Svenson (50:09)
I think the best thing that anybody can do, and I recommended this to you and you you've explored it, is a tool like Whisperflow. Now it doesn't have to be Whisperflow, but Whisperflow is the one that has emerged as the big one. but the idea that you speak words and those words get, converted to text and then you pump it into the AI. Because all these tools have their own little microphone button, but none of them are as good as

a tool like Whisperflow. I'm sure Whisperflow will have competitors. But here again I will cite Allie Miller because I was enthusiastically using Whisperflow thinking like I wonder why she's not recommending it because she's always, on top of these things. And one day she said, Whisperflow is the game changer. Mark Andreessen, same thing. So Whisperflow has the endorsement of a lot of people. I was lucky enough to have the founder, Tane Kothari, I think is his last name.

come and talk to the inner circle people because I was using it early on and he said, I wanna I'm happy to come talk to your lawyer group. We have lot of lawyers that are using it, a lot of big law firms, even back then, were using it. So it's a tool that makes a big difference. And if you think about why, it's because you need to get a lot of thoughts out of your head quickly. And it's gonna need to either translate audio into text, which, was one way, but it's better if you know what text it is interpreting, right?

look at it, make sure names are spelled correctly, what you said is what you need to say. So Whisperflow, huge game changer. I use it on my phone because a lot I think that's also important. If you can get work done on your phone, that's better. And the thing with AI is it can it doesn't matter how you give it the input, once you give it, it'll go off and do its thing. So I give it a lot of work by, logging in, giving it an instruction, telling it to go do something, but I couldn't give

it enough context as easily if I couldn't speak. So I if I were relegated to t

Jason Marsh (52:03)
So so just real quick,

the idea with WhisperFlow. And the funny thing about it is I had heard about it. I'm I'm checking it out. on the website, I come across there's a picture of Ernie with a quote on how great it is. So I started texting you about it. I'm like, look, see this guy, he's everywhere, he's getting ⁓ but but the idea there is that instead of typing everything, you're just talking into your mic and it's cool because obviously it takes what you said, but it fixes it and you could be like, scratch that last sentence and do this and create this list and

Ernie Svenson (52:16)
Mm-hmm. Yeah?

Yep. Yep.

Jason Marsh (52:31)
It it's really intuitive in a way that like just speaking directly into the like the chat GPT voice prompting maybe isn't quite as much. But yeah, super cool tool. I would check it out. Maybe we'll put a link or something in the bottom. But okay, so that's whisper flow. so start there. That's a great just efficiency tool. Anything else that you would say for an attorney like what they can start doing like

Ernie Svenson (52:49)
Mm-hmm.

Jason Marsh (52:55)
this month or maybe the biggest opportunity that somebody's currently not doing that they can start doing relatively easy.

Ernie Svenson (53:02)
I mean we've talked about a lot of these different things, but I think I I like to keep simple and actionable because if you give people too much stuff, then they don't do any of it. So I r really am gonna emphasize Whisperflow because that enables all the other things to fall into place more readily, right? Like that's an enabler of other things. And you know, beyond that one I would just say, whatever note taking tool you like. I mean there's b there's lots of them now. there's this tool called Plod, which I have like the device here.

It fits on the back of your phone. and you push a button, it'll start recording. So it helps you record conversations that you're having in physical space. But I also use it sometimes myself. It's just like I push the button, I walk around, I dictate. It's got 30 hours of recording time, so that's a lot. And I just brain dump ideas into it. And I think that's another area that's really useful. Just getting stuff out of your head. Again, it can be amorphous because the AI will make sense of it.

But what tool do you use? You can use your phone. The problem is then you're using a battery life on your phone. And this tool is one that I've noticed a lot of lawyers are they love the hell out of it. they like Whisperflow a lot too, but this plod device they also

Jason Marsh (54:14)
Yeah, it's it sounds great because what you just said, there's there's often a lot more friction than we realize. Getting ideas out of our head documented somewhere when where I gotta get my phone, I gotta open the app, I gotta find the app, I gotta f open the file, and then it whereas this it sounds like you could just record what you're saying and eventually it now it's at least there. It's it's been memorialized, and then you can find out where you wanna put it, and of course use AI to summarize it and make some magic with it. So, Ernie, this has been great, man. I think we're gonna wrap it up here.

Ernie Svenson (54:30)
Yep.

Jason Marsh (54:43)
I'm here with Ernie Svensson, Ernie the Attorney, as many people know, but we've been talking about how law firms can use AI to capture institutional knowledge, build better systems, reduce operational bottlenecks, and just ultimately run more efficient practices. And so you can learn more about Ernie and what he's up to at Ernie the Attorney dot net. and you can check out his inner circle. It's his community for solo and small firm lawyers.

Focused on AI, tech, automation, outsourcing, all things practice management. And as I said early, I think it functions almost like an R and D department for small firms where the members can test new stuff, share what's working, and just learn from each other's experiences. Is that is that a good depiction of what you're doing over there?

Ernie Svenson (55:27)
That's exactly

Yeah, that's exactly it. I mean it it is it is helpful to try these things, obviously, you should try them. But then the next best part is to go to a place a place where other people who are like you, other lawyers like you, are trying it and compare notes, right? Like that gives you more confidence, it helps you figure out what to pay attention to more. And so people inside there love that they can talk to other lawyers about their use of AI. And I I'm glad, even though I'm not practicing anymore because

I'm learning from them. Because in the end of the day, it's not what it's not so much about its use in the practice of law, although of course that's where we want to use it. It's really just how can it work generally? Like it applies across all businesses. The the knowledge of how to use it in any kind of business will help you in any other kind of business. lawyers, have other things they have to consider, ethics and confidentiality, and we talk a lot about that. So yeah, it's a good group.

Jason Marsh (56:20)
Yeah, and

I think that's great 'cause I think the idea there's a lot used to be, or there was, or are, training programs. I think that's less important now because you can get a lot of that information in the AI tools. What's more important is the community where you're able to learn and share, ask questions with other people and have these experiences. And so the last thing I would say about Ernie, Ernie's a great hang, right? As obviously, y you've seen that, but he does

Ernie Svenson (56:41)
Yeah.

Jason Marsh (56:44)
a really great job of taking complex ideas and making them understandable and simple as obviously you've been hearing today. So another thing you can do, subscribe to Ernie's newsletter. I do. It's a great quick read from around once or twice a week. And, he's writing about all the stuff we've talked about, AI and tech and, just essentially how to build a better raw practice. So I'm gonna include links, to all this stuff that we've been talking about in the show notes. Ernie, before we go, anything else you want to add that we haven't touched on?

Ernie Svenson (57:12)
No, just go out there and practice safe AI.

Jason Marsh (57:15)
Nice man. All right. Well, thank you so much for joining me today. Like I said, man, you're always a great hang. I've enjoyed this. And so this is Ernie Spence and everybody, Ernie the Attorney. That's gonna be it for today. Lots of great new content and shows coming. So stay tuned, subscribe, check it out. But we'll see you next time.

right, bye, everybody.

Jason Marsh (57:35)
Thanks for listening to the Alpha Lawyer podcast. I'm your host, Jason Marsh, and I appreciate you joining us today. If you enjoyed this episode, please subscribe, leave a review, and share it with other lawyers who might benefit. You can find more resources and insights on building a thriving law firm at alphalawyer.io or the Alpha Lawyer YouTube channel. Until next time, keep building the firm of your dreams so you can live your best life.

Jason Marsh (58:02)
right, bye, everybody. Already, man. Thanks, bud. That's ⁓ I'm gonna

Ernie Svenson (58:05)
Thank you. Yeah.