Agentic AI Is Reshaping the Legal Workday (Full Transcript)

[00:00:01] Speaker 1: Hi, I'm Zach. And I'm Stephanie. Welcome to the Lawyer's Podcast. One of the ways where we help lawyers build healthier firms, better businesses, and more sustainable lives. Today, Zach's talking to Sam Harden about the claudification of chat GPT.

[00:00:19] Speaker 2: The claudification of chat GPT. I love that. It happened. And we'll get into that here in a minute. But Stephanie, not a lot of people, I guess, would know this, but you and I like to talk soccer a lot. And the biggest soccer event in the world just happened. And if you didn't realize that the World Cup just happened, come out from under your rock at this point. I know.

[00:00:44] Speaker 1: It was sort of like, you and I talk soccer all the time. And now everybody got to talk soccer for a few months. It was- Come join our club. We do this all the time.

[00:00:54] Speaker 2: It was kind of fun to not be shunned out of the room when we were talking about it at the beginning of meetings. Yeah. It was neat. It was very neat. But this World Cup was, like every World Cup, had ups and downs and amazing things. But one of the things I think that's neat about something like this, Olympics, World Cup, whatever it is, is that you have these people who are experts. I mean, for lack of a better word, they're at the top of their game, literally. They're amazing. They get paid to do this. They spend all their time training and all that, and they're having to compete against each other. And emotions are running really, really high. And there's a lot of highs, a lot of lows. And I think you can see a lot about people's character in these games, in these tournaments.

[00:01:48] Speaker 1: Yeah. I mean, my husband was like, why do they even have a third-place game? This seems cruel, almost, to make England and France play for third place. But as you and I were talking about it, I mean, they showed up, and they fought, and they fought back. And it turned out to be an amazing match that was so fun to watch. And we also said it was sort of the American version of soccer, because it was high-scoring. So we would like that versus the final, where you're like, is anyone ever going to score?

[00:02:15] Speaker 2: Oh, man. Yeah. I don't want to talk too much about the final, because there's a lot going on in that one. But that also had a lot of character in the face of adversity. But that third-place game specifically, and various other games, yeah, both these teams lost their semifinal. And it would be really easy to say, eh, we'll show up and go through the motions. And I'm sure there are people out there that will argue that that's what they did. But if you watch the game, I mean, Jude Bellingham, he's been one of my favorite players for a while. And Jude Bellingham showed up and really kept fighting, and kept fighting, and kept fighting. And you see the class of this person that says, I just got beat, but I'm going to show up as a good, with my best self.

[00:02:58] Speaker 1: Yeah. And honestly, that extended beyond the players. I was in Chicago during one of the, I think it was during quarterfinals. And so I went to a watch party, because I was there by myself. Nice. And so I went to this. I Googled where to watch soccer in Chicago, and went to some venue with like 5,000 people. It was insane. And that was the match England won. And instantly, everybody just broke out into, hey Jude, singing, which was fun. But also, I went by myself, and I ended up joining this table of people. There was an empty chair. And I was like, hey, can I sit here? And they're like, yeah. And I made a new friend. We connected on LinkedIn. But there was just among the crowd, and honestly, among all the fans that came together, right? I loved watching all the videos of people who came to America and experienced America differently. Yeah. And all the silliness of refills on nachos, and drinks, and Buc-ee's. I mean, it was just a cool time. I think sometimes we get into all the things we get into, and it was just about people showing up and remembering that we're just people, and we all have these common goals, and lives, and shared values that don't exist. And at the end of the day, we're just people. And maybe even, you're probably like, our listeners are like, what the heck are Zach and Stephanie talking about today? But it's like with our profession, too. We fight like hell for our clients in court. But at the end of the day, we're all just people, and we all do have these shared values. So we got to show that same professionalism when we show up for our jobs.

[00:04:44] Speaker 2: Yeah. I like that. And I think that is, to me, the takeaway here is that it gives me heart. It makes my heart happy to see that class after adversity. And so in the same way, in everything that we do, having kind of class after adversity and recognizing the humanity of everybody else. Yeah. We could talk World Cup, because I also went to a watch party in Kansas City, because my truck broke down on a trip. That's a story for another time. Yeah.

[00:05:17] Speaker 1: We'll come back to that one.

[00:05:19] Speaker 2: But we could talk World Cup stories for a while. But we've got Sam on the podcast today talking about the claudification of chat GPT. And if you don't know what that is, stick around and listen to it. Sam is brilliant. He really is. So now here's my conversation with Sam.

[00:05:54] Speaker 3: Hi, I'm Sam. I am an innovation strategist at Affinity Consulting.

[00:05:59] Speaker 2: Hey, Sam. And welcome back to the Lawyer's Podcast. I know we said we were going to have you back again after the last time, because we were talking about Claude. And that actually turned into two episodes instead of one. And now I think I was like, we need to have Sam back on a lot more quickly than we expected, because man, it's AI. It's moving forward pretty quickly. I wanted to talk today about what, I don't know, I'm calling or other people are calling or whoever is calling the claudification of chat GPT. Chat GPT recently just put out. Is it chat GPT work?

[00:06:44] Speaker 3: Yeah. Yeah. What's going on with that? It's a confusing name.

[00:06:47] Speaker 2: Yeah. Everything is co-pilot or work or chat or whatever. Everything is that at this point. But yeah, so it's chat GPT work. And that, explain that to me.

[00:06:59] Speaker 3: So I think chat GPT has existed as like a back and forth chat app. I'm going to say a long time. It feels like a long time. It's really like two, two and a half years. I think in AI time, that's like two decades.

[00:07:14] Speaker 2: Since time immemorial, since memory runneth not to the contrary.

[00:07:18] Speaker 3: The beginning of time. I think they, and again, I don't, I don't know what is in their minds, but it seems like they were watching the Claude people with Claude code and Claude co-work and seeing what they were doing and said, we need to get in on that. And they have released a desk, a desktop app called chat, chat GPT work slash chat GPT codex. That's like one app, right? That does everything, but has like five names for anything that it does. Yeah. It's, it's, it's interesting. It's, it's a lot like the Claude desktop app with Claude co-work and Claude code all bundled into one system. But now in the chat GPT flavor of things.

[00:08:16] Speaker 2: Let's just in case people on the podcast don't know what we're talking about with this kind of like switch from, you know, like chat GPT has been a ask an answer sort of thing for a while. And I think a lot of people have hacked it together to say, ask an answer. And then it really helped them, you know, vibe code and stuff like that. So they've gone into some really, really complex prompting and, and things like that. People got really, really good at this. Ask an answer. What is chat GPT work in relationship to that? How does this change literally what, what the, I guess, interface even is.

[00:08:57] Speaker 3: So it takes chat GPT like that back and forth discussion, right? Where you ask chat GPT, well, help me answer this question. Help me do this thing. Help me figure this out. And you have a back and forth discussion where now you're able to give chat GPT its own little workspace that it can use and the files it can use in there. And it can do what developers and computer people call CRUD, C-R-U-D, create, read, update, and delete. If I remember correctly, I always get something wrong in that.

[00:09:30] Speaker 2: Read is the one that I always forget. Yeah.

[00:09:33] Speaker 3: Yeah. I want to say write, but write doesn't start with an R. No, no, it doesn't. But it gives it that CRUD ability to manipulate files in that folder. And what that lets it do is instead of just having a back and forth discussion, you can now have chat GPT work, whatever they're calling it. Go in and use a bunch of files at the same time and make updates to the files and sort of act more like really an agent on those files and take actions and make plans and things like that. Last year, people were talking big about agentic AI, but when you looked under the hood, it was things like, well, you can create a custom GPT that goes and checks Reddit or looks at news articles and things like that. But now when we talk about agentic AI, we're talking about the real capability to do things on files and to do things in the real world almost.

[00:10:37] Speaker 2: To take action. And I think that the word agent and how we as lawyers think about it actually makes more sense in this area. They are acting kind of on your behalf in some ways. So chat GPT just released this work. It looks, feels, sounds a lot like Claude, Claude co-work and the direction that's going. So what is, I guess, what is the shift here? What does that tell us? Why are they going? Are they just like, hey, we're going to, you know, Anthropic seems to be doing well, let's do what they do. Why are they releasing that also here?

[00:11:21] Speaker 3: I think they, one, I do think they're looking at Anthropic and what Anthropic has done with co-work. And I, like Anthropic hasn't really said, here are, you know, they're not a publicly traded company. We don't, you know, we only have so much vision into what they're doing. But it does seem like co-work is a big success for them. And it was kind of an unexpected success. They have talked about co-work being something that they created because office workers and white collar workers were using Claude code to do things like create documents and manage big groups of data files and things like that. So not programming, but like HR departments using Claude code to go in and manage a ton of records or things like that. So open AI is looking at that and saying, well, we have the capability and we want to give people an alternative system to be able to do the same thing using our models, using our platform, paying us instead of paying Anthropic, right? I mean, sure. But I think, you know, they have a ton of users using chat. They have a lot of people using the free version or the $20 a month version. And I think they see the potential of this kind of agentic app for workers or for just anybody who wants to manage things to give them that capability. And I think that's kind of what they're seeing. That's their motivation there.

[00:13:02] Speaker 2: Well, and so we kind of have to assume that Microsoft Copilot inside of Microsoft 365 is going to go the same direction, right? And that Jim and I will go, you know, we'll at least have something that is related to this. And so, yes, the entry point to this specific episode of the podcast is the claudification of ChatGPT. But really, we probably need to think about this kind of like platform agnostic. And so with basically any of these tools, what does this do? Like, what is the mindset shift? Because that's what we're talking about. We're going from ask and answer to define. And you and I harp on SOPs all the time, define, define how something is done, and then delegate. Really? I mean, because that's what agentic AI really is, is us delegating stuff. What does that do for law firms?

[00:14:02] Speaker 3: I want to take a step back because I think the mindset shift thing is really foundational here. To borrow a claudism, like it's, you know, Claude will say that's foundational. Now that's in my vocabulary. It really does take a mindset shift because we have gotten used to the idea of AI as a chatbot, where you, you know, you ask it questions like a lawyer takes a deposition or if you remember, I don't know if people remember phoning libraries to ask a question.

[00:14:37] Speaker 2: I do. Right? You know? I would do that from the bar way too much. Yeah. Because we'd have some sort of bet. And I'd be like, who was, who did do this? And then, sorry, this is going to be weird, or a weird tangent. There's the movie, The Desk Set, that I always think about with, it's Katherine Hepburn and Spencer Tracy. And it's when computers essentially came out and started taking over for the in-house librarian for like the company. But yeah, when you could phone, I remember phoning into libraries for me, it was Ole Miss in Oxford. Yeah. Sorry. Weird tangent. Weird tangent. Yeah. Keep going. Yeah.

[00:15:21] Speaker 3: But yeah, it's like, we have gotten used to that idea. And now we're looking at a system where you can literally give it, if you're working in a case and you have five depositions, a whole stack of pleadings and motions and some research materials that you've done, some records, some, you know, case notes, whatever you have in there. And you can hand it off to an AI system and say, not just what exact date that this thing happened on, but go in this case, analyze these records this way, outline a plan that involves these steps and then save it because I want to see it. I want to use it. Or go in and look at all of the depositions in this case that I've done and outline what you think, AI, the important pieces for trial testimony should be. Now, you shouldn't automatically just go with what the AI says, obviously, but it's capable of doing that kind of flow and reading the files, writing files, doing its own analysis, all of that stuff. It's able to connect with online case databases now and go in and say, well, I need to get a better handle on promissory estoppel because nobody understands that. And I'm going to go research that, or I'm going to go look at what does the law actually say on the statute of limitations for this kind of matter when the notice happened within this time period, because that's an issue, right? And make its own sort of mini decisions along the way. That's the kind of thing that we have now that is still being built, but we're still in the mindset of, well, I can go to chat GPT and ask it a question like, what is the law on writs of replevant, or promissory estoppel, or remitter, or whatever fancy legal term you want, and sometimes it gets it right, sometimes it gets it wrong. That's where we're at. But in reality, I think the technology has shifted to being more of this agentic coworker that's capable of planning and executing and creating.

[00:17:44] Speaker 2: And I think this is the next step in my brain. This is where I go in trying to get people to shift the paradigm. When we have this agentic ability to go and connect to disparate data stores, let's say. So one of your data stores is going to be your practice management system. One of your data stores is going to be the file that you're working in for Claude Cowork or chat GPT work, the file folder that you're working in. And when I have the ability to have this agentic platform, this LLM, connect to five disparate platforms, now it becomes my working interface instead of me going into those places. So I don't have to go into the Word document that has the deposition. I can essentially ask this Claude or chat GPT work. They need to come up with a better word for that because it's so much easier to say Claude or mod chat GPT work. I can ask it, but I can ask it that with a query essentially to another data repository or another app or something like that that I'm connecting to. So it becomes this interface because of its agentic ability to go do these things as well.

[00:19:10] Speaker 3: There is, if there's not already, there is going to be a conflict over what the surface of work looks like. This is how I think of it. This is kind of an out there idea, but when I was practicing... Let's get out there. Yeah. Huh? Let's get out there. I'm with you. Yeah. Yeah. I can't remember the things that I use at a time entry software that I cannot remember the name of, but it was where I put in time. I had Microsoft Word open. I had Chrome open with Westlaw, with the Florida statutes, with the Florida administrative code, all of these research materials in different Chrome tabs or in Chrome windows. I had Outlook open. I had maybe Excel. I had a bunch of different surfaces open based on what I was doing, and I think a lot of lawyers are the same way in that they have all of this stuff out there. They have seven Word documents open. They have a PDF reader open with a bunch of PDFs. They have Westlaw or Lexis open or Clio Work open and whatever research platform they're using. They have Outlook. It's like, and they're constantly context switching between all of these platforms, and it's almost like ADD supercharged every day when they are working. What we're seeing now is the consolidation of all of that stuff into one interface, and I think that's becoming more and more possible where you can literally go in Claude or in Chat GP to work and say, what's going on in the Smith v. Jones matter? It's able to go and look at your email, go and look at your calendar, go and look in the case file materials, maybe even check the court docket, and pull all of that information in and show it to you without you having to think, I can't remember what's going on in the Smith v. Jones matter. First I need to go and read, I need to do a search in Outlook, and then I need to look at, I think there's a calendar event, but I can't remember what it's called. Let me go and figure out where that is. Oh wait, I think the judge said something in this hearing that we just had last week that I can't remember, but I know it's in the hearing transcript. Let me go and see if we have that transcript in yet, and then find it. Spending all of that time and spinning your wheels, the AI systems now consolidate that into one surface, which I think is really, really nice, but it makes me wonder about what work looks like in other platforms five years from now. Will we be writing things in Word? I don't know.

[00:22:00] Speaker 2: For me, and I'm going to pick on Clio here just because I think they're doing a really good job of staying out ahead of the curve, so I think they can be picked on. My father always said, punch up, not punch down. Pick on Clio here, and I don't think I'm saying anything bad, but am I going to log in to Clio in the future? Right now I think, again, they're doing a good job of having Clio work to where they are starting to own that surface, but am I going to log into my email? I think about this in interaction with websites right now. I think of it as bullshit that I have to go and click on things in a website. Why do I go to the website and click? Why wouldn't I go to Claude? For me, I actually work in GitHub Copilot. My work surface is VS Code, and so I'm in GitHub Copilot. I build MCPs, and MCPs are just a way. It's model context protocol. It's a way for these models to interact with anything that has an API, essentially. So I build MCPs into the platforms that I want to talk to, and that's where I do work. I don't open my email. We use Teamwork. I don't open my project management platform. What am I supposed to do today? How does that affect the other stuff that I want to do? Getting out there even further, this isn't going to be this way for that long because eventually that surface is going to become my phone, which we can do right now. We can have that surface be your phone, but it requires a basic paradigm shift for attorneys to think about how they work in a fundamentally different way. Instead of ask and answer, summarize all that stuff that we were doing six, eight, ten months ago, heck, you might be doing it last week. We got to shift that. We got to change that because there's some real amazing stuff we can do in a law firm. I think you've gone over some of the ideas of what we can do. Walk me through that in a firm. Walk me through what an attorney can really be doing, some of the things that you've been playing with in your claw firm and things like that.

[00:24:34] Speaker 3: I want to make one note on this, kind of on this tangent, is that I listened to a podcast recently with Boris Cherny, I think I'm saying his last name right, who is the guy who created Claude Cote. So if you think about in the pantheon of computer developers, the guy who wrote or created the AI system that is the default AI system for better or worse that computer programmers use now, literally said something to the effect of, I know that's a contradiction, but whatever.

[00:25:10] Speaker 2: He said exactly this. Exactly this, kind of, in paraphrase.

[00:25:18] Speaker 3: He was asked, how are you working with Claude? And he said, well, I do it on my phone. I interact with Claude on my phone. I just have Claude do things. I just tell Claude from my phone what I want him to do.

[00:25:36] Speaker 2: And I ask him what I need. Yeah, hey, Claude, can you do this for me?

[00:25:40] Speaker 3: So I think, yeah, the future of what we have is going to be different from what we have now. I don't know what that looks like. But when I think about law firms, and I think about the way that cases move and the documents and the systems that law firms have in place, I really think a lot of it is right for an agentic system to come in and work on with the attorney. Now, the key word there being with, not instead of, not in place of, because I don't think that we want a world where AI is making legal decisions for a law firm without the attorney being like the wrangler or the ringmaster or the, you know, whatever the term is.

[00:26:35] Speaker 2: Yeah, even if you can, I mean, like, just that's not what we want as a society.

[00:26:41] Speaker 3: Yeah, law is human, right? Our courts and our legal system should be a human process. But I think that agentic systems can take over a lot of the mundane busywork that lawyers have gotten so used to doing, things like opening files and creating the little structures that go on inside files, like creating strategies that are pre-basic and saying, okay, I need to figure out who I need to depose, go and look at the, like the accident report or look at the witness list from the other side, you know, help me plan that out. And taking away a lot of that, like just wrote, copying and pasting and writing. I think about the way a lot of attorneys work now, even with the availability of document automation, right? Where lawyers will say, I want to look at document automation, but the way I do things right now is I go and search SharePoint and whatever DMS, document management system I have for the last time I did that. And I copy it over and then I change things in there. There's no templating, no standardization, and everything becomes bespoke by default, right? Not because it needs to be bespoke, because it's the motion you file every time this happens, it always has the same, you know, like jurisdictional language. They kind of make it bespoke by default. But taking things like that and being able to say, hey, Claude, hey, chat, GPT, the next step is to create this motion, start the process. And it's able to use a set of templates and go in and do that for the attorney without the attorney having to spend their time going in and finding it and then manipulating things and doing all of that stuff. And it becomes almost like a junior co-worker, for lack of a better term, where you're able to give it tasks, it can run, even run on a schedule, all of that stuff. And the attorney takes a step up and is able to look at what the AIs are doing in each matter, in each department, and be that, like, upper-level management and operate at the top of the license, not being a scrivener, you know, fixing things in Microsoft Word all day.

[00:29:20] Speaker 2: Sorry. I love that as an example, because that's what we do, is we reformat stuff and we make it look pretty and get it redlined. Like, that's a significant portion of what, like, a lot of attorneys, I think, have considered themselves Microsoft Word power users, because they're really good at that. And I don't know that that many people really like doing that, specifically, as much as they would like designing the system around how do we, how do we review contracts? How do we, like, what clauses do we put, like, I'm great with writing the clauses, you know, like, the exact clause that you want. Well, then it goes into some sort of clause repository in Claude. And then you can, you know, use that. But I really like how you talk about operating at the top of your license here, because we, I think a lot of people would hear this conversation and go, I don't understand it, and I don't want to understand it, because all it's going to do is make me, it's going to leave me with a crappy part of my job. You know, we want AI to think for us. No, I want AI to do the bullshit. I want it to, like, I don't want to have templates. I want to have the idea of how I write something. And yeah, we can get into some real specifics of exactly how we write things, but I want to have the idea of how I write something and then be able to say, write a letter, you know, using, go and get the facts from this file folder and write a letter to the attorney on the other side. And you could figure out who that is in this file folder.

[00:31:18] Speaker 3: Yeah. I don't want to have to go hunt for their email address. I don't want to have to hunt for their address.

[00:31:25] Speaker 1: Yeah.

[00:31:25] Speaker 3: And I think a personal pet peeve of mine, just to expose my pet peeves, is when legal technology demos veer into the territory of AI is amazing. We're going to have it brief this Supreme Court argument. We're going to have it do this really complex, heavy lifting reasoning task. And that's what they start out with. And I really think AI is very capable of doing a good bit of that in a way. But I think that a lot of lawyers, including me, see that and think, so is this future that we're looking at one where I am the scrivener and I am just giving Claude Fable, you know, GBT, whatever, here are your materials and I will be your scrivener and I will make sure the formatting in Word looks right for your incredible arguments, Mr. or Mrs. AI, whatever. I don't want that to be the future. I want it to do the bullshit work that I don't want to do, right? I want it to do the formatting. I want to be, because I went to law school, I spent a lot of time figuring things out, practicing law. I want to be the one figuring out, how do I talk to this witness? How do I present this argument in a way that's persuasive, but powerful and yet compassionate? How do I balance those three considerations? How do I connect with this witness or with jury or with this client? How do I present this in mediation so that, you know, we can reach an appropriate compromise and get my client what they need? I want to be doing that. I don't want to be sitting back in my office going, wow, Claude, you really wrote an incredible mediation statement that needs no changes, then I will format it. Right.

[00:33:32] Speaker 2: Now, let me fix the formatting because you're shit at that.

[00:33:36] Speaker 3: Yeah. I think we've, I worry that we're looking at it backwards a lot of the time. And I want the future to be one where the attorneys are up a level, watching the AIs work on different things, but being the ones making those decisions of what needs to go into this. What is the appropriate plan of attack? What is the set of guidelines that we're going to apply to this contract so that we negotiate it?

[00:34:07] Speaker 2: Well, that's where I think things should be going. I like that rather than me having to figure out where the freaking address of opposing counsel is. Yeah. Okay. So you and I could wax philosophical on this for forever. And we do in our, in our team's chats. But one of the things I think we've, we've gotten to is that there is an entry level issue with people not knowing what these platforms can do. What would you suggest somebody does if they're just like, I'm getting started with Claude, I'm getting started with chat, GBT, um, work, what, what should they do to figure out honestly what they can do?

[00:34:53] Speaker 3: I think the best method right now is really, and truly to do a little bit of experimentation.

[00:35:00] Speaker 2: Okay.

[00:35:00] Speaker 3: So I think keeping in mind considerations around confidentiality, if you're using a personal account, but if you have a personal account and you're using it for personal things, see if you can get the, the chat GBT app or the Claude co-work app on your computer, download it and go in and just maybe give it some, some old cases that you've looked at, right? Like appellate court cases that are public record, not from your cases. Gotcha. Not my old cases. Not your old cases that I've gotcha. Okay.

[00:35:34] Speaker 2: From public data.

[00:35:36] Speaker 3: Yeah. And see what it can do around those. See like, Hey, I want you to create an outline of the arguments that were presented in each of these five cases, um, format it this way. Um, and just kind of play around with it and see what it can do. Take a look at the connectors that are in each Claude has a ton of connectors, chat, GBT work, whatever they're calling it, has a growing list of connectors to legal research platforms to things like DocuSign to Google drive to box to, you know, all of this stuff and start thinking through like, well, what if I could connect it to my Google drive and I have these spreadsheets in there that have this, could I ask it to go in there and, and pull this out and format it this way? Could I ask it to do this? Could I ask it to do that? The answer to, can you ask it to do something is always yes. Whether or not it can do it is you just have to ask it.

[00:36:41] Speaker 2: Oh, I love that. Yeah, I can ask it. Absolutely. Yeah.

[00:36:48] Speaker 3: Oh man. That's the best way to find out. Can it do this? I don't know. Have you asked it? Have you asked it to do that?

[00:36:58] Speaker 2: I like that. Okay. So get one of these things, ask it, ask it if you can do the thing that you dream of. And I have found that as I learn more things that these platforms can do, I can dream of more stuff that these platforms can do. So get out there and kick them around. Kick the crap out of Claude.

[00:37:20] Speaker 3: That's yeah. Tell Claude what you want. I'll chat GPT. I want this. Can you do it?

[00:37:25] Speaker 2: See what it does. I love that. I love that. Awesome. Well, Sam, I'm sure we'll have you back on very soon again, and people can always follow you on the LinkedIn and you've got a sub stack that I always like promoting. What is that sub stack? And we'll put the link in the show notes.

[00:37:44] Speaker 3: It's called team do something. I yeah, team do something. It's an old sub stack that I still write on about once a month at this point.

[00:37:52] Speaker 2: Well, it's always good stuff. I enjoy reading your sub stack, Sam. Thanks for being with me and talking about what do we call this agentic AI and the surface of work?

[00:38:05] Speaker 3: That sounds really fancy. Sounds like we know what we're talking.

[00:38:10] Speaker 2: It sounds like we know we're talking about. I don't. I don't know about you. Claude. I've had Claude up the entire time.

[00:38:18] Speaker 3: I really I promise you, I have not.

[00:38:20] Speaker 2: This is just I gave Claude a sample, a two minute sample of my voice, and that's just what this is. Is Claude talking in my voice?

[00:38:27] Speaker 3: Yeah.

[00:38:28] Speaker 2: Awesome. All right, Sam. Well, thank you once again, and we'll talk to you next time. All right. Thank you.