OpenClaw: Your AI Assistant That Actually Gets Things Done

**[00:00:00]** Hello everyone, my name is Evando Prescaro and welcome to the web talk show. Today we are talking about Open Claw, formerly known as Maltbot, formerly known as Claudebot. And this name changing has occurred only in the past couple of days. So it's very funny if you are keeping track of it. It's been a very interesting roller coaster ride, especially for the author of the tool. But we're gonna talk about what it is, what's it about, is it worth it, is it just hype, what happened with the crypto pump and dump, etc. It'll be a fun episode. So stay tuned if you've heard anything about Moldbut, Cloudbot, or OpenClaw, and are interested to see if it would make sense for you to use it. So let's talk about what it is, what it means. First off, if you've been around the LLM industry, the LLM space, the AI space, you probably have seen that the interactions with said LLMs is typically through a sort of chat interface, right? Might be chat interface, might be voice interface, and that's how we talk with the LLMs. Then that's cool. Now along comes Anthropic and then others followed and they create code or the CLI version that you can then use from a terminal. This is more sort of developer focused. It's pretty neat because it allows you to use the terminal and talk to the tool to the LM through the terminal. Little technical here but I'm just explaining some background. Now that is great because it has access to files. So it can talk to the computer, it can create files, it can create folders, do all sorts of things that makes it way more powerful than just having it in the browser or in the app. **[00:02:00]** Now developers then started being creative and said well the thing is we are using it but then we ask it to do something and it's more of a turnbyturn things. It will do something maybe even think about it depending on the model you're using. It'll think about it a little it'll do some things and then it'll sort of stop. So it can do many things but it doesn't continue a very long running task of just finishing up a fullfledged project. Now, of course, again, Entropic OpenAI with Codex and other tools, they started implementing ways for you to dangerously skip permissions and things like that so that it can actually run autonomously and do many things and so it can spawn sub agents or additional tasks or additional functions that are running behind the scenes and they're doing all sorts of fun stuff behind the scenes while the main agent is talking with you or interacting with you or even just keeping track of tasks. But if you've used it at all for the most people, it the experience won't be like what you hear in the PR releases or in the media where it's like, "Yeah, I left it running 30 hours a night. It just did everything for me." Not typically. Like it'll do one or two things and then it sort of will stop. So then some people come along and create a thing called the Wikum the Ralph Wigum loop. And it's very interesting because it allows you to feed it a PRD or a product document that explains exactly what the tool should do. If you're building a tool, if you're building a software or a website or whatever, you can give it this robust document and it includes tasks of **[00:04:00]** what it's going to do. And then what the wig loop does is essentially running a loop of I have this task. I'll send it to the CLI to the cloud code or whatever you're using. And then once it's done, I'll test it and then I'll go to the next one and continue. So basically, it's it's not like cloud code is doing everything or or codeex or whatever. It's this other tool that's a loop basically that's just going through each task and then submitting a request for each of those completing them and then as it proceeds gets a lot done. And that's amazing too. We've tested it. It works out really nicely. You're able to do things on its own overnight etc. There's of course the inconvenience of perhaps it might use a lot of tokens etc. And so that depends on what plan you're using many things that you can ask me any questions you have about this and I'll be happy to talk about it but one of the things that I personally disliked about the loop just like a whim loop or anything like that is that you lose some visibility. So it just sort of runs unless you create another dashboard for it so that it runs and you can see what it's doing. And it's not so easy to stop things and stop midway and do some other things and it makes a lot of decisions on its own, which is okay because it's supposed to be autonomous. But I think that for many use cases, you want to be in the flow a little bit. You want it to be able to to tell it, I don't have this huge document ready for you. Let's just chat about it, but **[00:06:00]** then go and do the whole thing. So have this sort of interactive experience and then allow it to do that. So then comes Claudebot and I don't remember what the original name was. I think CL or something originally and then it became Cloudbot got too much press because people were saying it's the be all end all for software development. It just does everything for you. It's an AI assistant and there's nothing like this that has ever come to the face of this earth and it's changed my life, etc. And yes, it's an amazing tool and I'll go into what it actually does. But then there's there's two sides. There's one side that says, "Oh, this is just a glorified rapper on the thing, and there's other sides that say it's miraculous." So, I'll explain what it is and how you interact with it and why I think it's brilliant and it might not be what stays. There might be some other things that come along, but the project I think is a fantastic idea and we've been using it and the way we've been using it collaboratively has been very useful. I'll talk about a little bit of that as well. What is it? Well, you can with what is it called now? Open claw. You can go to their website and I think it's openclaw.ai and it used to be moldbot like last night or something the day before and you'll see what it is. This will explain it. Basically, it says it runs on your machine. You can talk via any chat app. So, you can talk to it via WhatsApp, Telegram, Signal, Slack, iMessage, Google Chat, etc. And it works in DMs and it works in group chats. So, I'll talk about **[00:08:00]** that as well. It has persistent memory, so it keeps track of your longunning conversations. And then it can control your browser. It has full system access like I was talking about with cloud code. And then it can also have skills and plugins. Again, if you've been you've been hearing anything about the space, you can have these skills that are sort of pre-built instructions that tell the tool how to use other tools easier without you having to explain to it how to do it. So, it's really really interesting. What I particularly like very much is that it's out of the box. Again, I'll explain. A few weeks ago, I released a video where I showed that we have our own version of Cloud Code running via Telegram in a server. So we have a server and we have cloud code on the server, but then we have an NA10 automation that allows us to talk to a telegram bot and just give it commands and vi10 we could do whatever we want. So it's really neat. It's like a wrapper of cloud coding in telegram and I liked it but I had to create the whole thing. I had to see okay how am I going to connect this and do the SSH and it's very technical and while it works and it's good and we use it for other things I really like that open claw does everything for you. The premise, I think, is if you have a computer lying around like this computer that's right behind me, it's an old MacBook Pro 2012 or something. You can have a Windows computer, a Linux computer, it could be whatever. It doesn't have to be too powerful. You can go ahead and buy a new **[00:10:00]** computer if you want. That's some people are buying Mac Minis. It's not that you need a Mac Mini. Everyone talks about the Mac Mini. The reason I think many people talk about the Mac Mini is because it's very easy to buy. It's just a little box. There's not a lot of confusion and it just works and it's very powerful in itself. small, doesn't consume a lot of power, and again, it just works, and it's not expensive. One of the least expensive computers that actually has power that you can get. So, I think that's why many people mention that one. Regardless, I think it's a great opportunity to reuse some of the older hardware that you might have lying around, especially like laptops and things like that. So, what I did, I just installed it on a really old computer just to test and see how it worked. The install process is very simple. that computer. I didn't have anything on it, so I just ran the command and it installed everything. In this particular case, I was installing it on Windows WSL, which means Linux on top of Windows. I decided to do that for other reasons, but you can do it directly on the Mac, and it's great because there's a lot of tooling with the Mac as well, as long as you have the right operating system. I think uh 13 Mac OS, yeah, I don't remember which version, but there's there's a version, a minimum version that it'll work on. Regardless, you can run it on Linux and it'll run easily. So, what's really nice is that you just run a command in your terminal and it will install everything it needs for you and then it will allow you to connect **[00:12:00]** your different tools. But all of that is already built in. And I think that's where it's really powerful. It allows someone who's not necessarily too technical to go in and install it. Now, here is the warning. Don't just go ahead and do this on a computer that has all your files or just go ahead and connect it to your email and everything where you have all your emails or your business accounts. Don't do that if you don't know what you're doing because you are essentially giving the LLM and this this tool access to everything. So, be careful in that sense. Let's be honest like you you really need to use your mind a little. It's very fun. But just use an old computer that doesn't have anything on it and and connect it to test accounts that you have, maybe another Gmail or something just to test it out and and get a better understanding. But if you are technical like we are and you are going to use it for certain things, it is tremendously powerful and I've talked to various people that have installed it and used it and we're all blown away with the ease of use of it once you have it running. That's what I really like that while I can make something like that in a couple of hours, the fact that it already connects to all the tools, not just Telegram, but WhatsApp and Discord and Slack and Signal and iMessage and everything just saves you so much time. It already works. It already accepts images. It already accepts attachments. It already accepts voice notes. And there's a lot it can do. It's very powerful. Now, why is it so important to have something like this? Well, it's **[00:14:00]** not important for everyone. Most people I would say are fine with having just a chatbt or a cloud or something on their phone and just chat with it. That's perfectly fine because you could do a lot with it especially how they've evolved even the co-work feature in anthropic as well. But what I really like about this apart from the ease of installing is that the long running thread or the long running conversations allows you to use it more as an assistant. So I think the main benefits are that you can talk to it via your chatting app. you don't have to be on your computer and still have the power of being able to create files and generate scripts and run applications and build full-fledged applications and run them and test them and everything which we could do regardless I know but that's one of the pluses another one of the pluses is that it allows you to choose your model so you're not like just just only open AAI or only anthropic you can there's multiple models you can choose from and you can connect to your subscription so you can use your an API token if you want and it'll usage based or if you have for example OPUS you have OKCloud what's it called max the biggest Mac plan well it's going to be $200 a month yes but you'll have a lot of space to work on right so if you're a developer you're in the developer team you probably have either pro max or max whatever it's called the biggest one and that way you're not constantly thinking of hey I'm running out of tokens but here's the catch I when I started playing with it I realized that it's **[00:16:00]** all in the tooling so it can change its own settings. What that means is you could if you wanted to tell it within its rules to say use set 4.5 for most conversation so we don't use so many tokens and then when I ask you to these types of things change and use the other model so it could use more powerful models for things that make sense for it to use. So depending on your allocation of tokens or how your subscription works, you can do things like that because again it can it can switch over and then there's so many things to talk about like you can have it talk to more than one person too. That's another great thing. So you have this thing that you can now chat with it from your phone or from your browser or whatever wherever you are and it's able to create files, run applications, install software etc. So very powerful just in that sense but then it has that persistent memory. So what that means is that I can be talking to it about something, right? And then say, "Oh, by the way, can you draft this email for this other person? Just keep it over there. I'll ask you for it later." And it could go ahead and do that in a sub agent. And then you continue with whatever you're talking about. Then you can say, "Hey, by the way, can can you send that email now? I think it's a good time to send it." And it'll go ahead and send it because it has this long lived conversation. Again, since it is using an LM behind the scenes, there's context to consider and sometimes it might forget some things, but it has a **[00:18:00]** separate persistent memory that allows it to keep track of a lot of things around how you interact with it, around how you want it to act, and many other things. So, it feels more like an assistant than many of the LLMs in my opinion and what I've heard from other people as well. It doesn't feel like give you an example when I use plot code that I use very often every day many times a day I find myself having to clear the session because of context because it's becoming a little less intelligent when the context window is filled but then even if I gave it very specific instructions in cloud MD which is it file sort of where it keeps track of what that project you're working on is I I find myself having to remind it many times this is what we were doing or this other thing and be very specific and so you have to babysit it a lot in my experience. It might be very different for different people depending on how you work with these tools. But in that sense, I feel that having OpenClaw has allowed me to keep that conversation go further and keep many conversations going. And that's the main difference because in cloud code you're typically talking or you have a session that's linked to a project and if you move to another folder another repository then you have different sets of conversations regarding that repository and yes you can have it talk to different files on your system etc. But it's not the most practical thing when task switching while having this sort of open claw cloudbot whatever you want to call it experience allows you to be all over the place which for a **[00:20:00]** lot of people that's how you work. you're working on two projects at the same time and you're doing different things. Maybe you're doing outreach, but at the same time you're working on client work and at the same time you're doing marketing and at the same time you need to do research, right? So having something like this, I'll explain what I did the very first day and what it solved for me and this might give you more examples of like real real life usage. And again, if you have any questions, please drop them in the chat and we can talk about it. So the very first thing that I installed after I installed it was I said, "Okay, so let's try this out." Because people say, "Hey, it's it's building things for me overnight and it's amazing." Yeah, sort of. But it can, but you don't necessarily have to be a constant builder of applications to get take advantage of this tool. It could also help you with repetitive basic tasks that are very important but basic. For example, I might want to know what is trending. I might want to know which videos came out this day. And I might want to do that research overnight. may want to know what happened during the past day, everything that came out. I might want to know which topics are being discussed on Reddit and blogs and other news sites. And I might want to know which videos from the creators or pages or topics that I like are popping off. They're they're just having a lot of reach and engagement. So, I might want to know about them because maybe I want to do a video or have my take on that as well. If I **[00:22:00]** do it manually, then I have to sit down and have to go over to the different channels and and do that whole thing. If I don't want to do it manually, then might maybe if I know what I'm doing, I might use NA10 and create an automation that goes and finds the channels and parses them and uses ampify or uses firecrawl or uses whatever uses different APIs and gets all the information. Again, I have to build that automation and that's what we would do many times for this particular situation. With this you can say and you can do this with cloud code as well but but here the experience is a little different. You can say, "Hey, James, Erica, whatever you call your agent, I want you to find the pages, channels, subreddits that make sense for my industry." Because maybe you already told it a little bit about yourself, and I go and find them, give them back, and you can say, "Okay, perfect. So, now that we have this, I would like you to prepare a report overnight and send me a notification in the morning at 7:00 a.m. and just go through all the channels, find the videos that have the most views in each of these channels, and then see which ones are popping off, why? What's the similarity between them? Are there any interrelated themes? What are people talking about right now? And what are the thumbnails looking like? What are the titles looking like? What is the engagement type? what length video lengths are working more today, things of that nature. So, you can just give it that, just talk to it, tell it, and then have it go off. And in the morning, you receive, and I did **[00:24:00]** receive a notification that said, "Hey, Rand, here's your report with all the information, and here's a list of videos that you can do. Here's some shorts, here's some longs, um, the cooks that you can use. This is what I learned about these videos that did so well, etc." So, it create the whole report, just send it to you in the morning. And that's where I think it's really powerful. the cron ability which cron is very simple. It's just a job thing that works on systems has existed since forever and it's basically something that can run during the day multiple times and just by it having that makes it supremely powerful because you can tell it do this at this time every day. So now I didn't have to go and create an automation. I didn't have to figure out how the APIs work. No, I just told it to do that. And if it needed to do something, it's really neat because it has this ability to find skills that are common that other people are using or to find MCPs that allow it to understand how other tools work. And there's a lot of videos on this. I I won't go into it, but it's really neat because it can go and install these MCPS. I can tell it, oh, I need to connect to Superbase. Can you set up the NCP? And it's like, yes, of course, I found it. I have it. I'm done. Or, oh, I want to talk to N10. And we did this yesterday. Can you connect to it? Oh, yeah, for sure. I just need the API key set up. Ready? It's done. So I didn't have to go do the documentation, figure out how it works, get **[00:26:00]** the API key, then spend an hour figuring out how it worked. No, it'll just do it and I can install tooling for it as well. So I did that. That was the first thing I asked it to do for the next morning. I got it. I was excellent. Yes. So I went and recorded the video. Then I said, "Okay, so can you transcribe this video?" So I sent it the video as an attachment via Telegram and I got it and said, "I got the video. It's 50 seconds long, but I can't transcribe it." But do you have an OpenAI API key? Because I could use Whisper for it. So Whisper is open AI's version of transcription type thing. And I didn't want to use my subscription for that. So I said, "No, I I don't have an API key for that. What other options do I have?" And I said, "Well, we also have an option of using OpenAI's Whisper open source version, the local version that doesn't need an API key. It just runs on your CPU or GPU. It's slower, but it works." I said, "Sure, let's do that." So it installed, went ahead, got all the packages, installed it. very old computer again running in like a virtual environment and read the video, transcribed it, gave me the transcript back and I said, "Cool. Now, can you find the golden nugget within that video and cut it for me?" And so it went in, read the transcript, said, "Oh, this part is really good from this amount of seconds to this amount of seconds. I can install FFmpeg, another tool, Linux Unix type tool that works with videos. A lot of video tooling just uses this behind the scenes. So, it installs **[00:28:00]** FFmpeg. Then it goes and cuts the video in that specific spot. Then it sends me the video back via Telegram. That's where the power is. I think it's making it accessible to just do whatever you need at that moment and not get in the way. I think that's what really makes it powerful. Yes, it's a wrapper that allows you to chat with it via your favorite apps. Cool. That's nice. But and yes, it allows you to use cloud code as a separate thing without being in the computer. Yes, that's cool. Yes, it has long to live memory. Okay, that's cool. Yes, it has crown so it can run tasks repeatedly throughout many days, every night, do a report, summaries, etc. Okay, that's cool as well. But the ease to which with it connects to other tools and installs what it needs, that's where it's again really super super valuable. And I will explain what we're doing with it right now as well because it's been mind-blowing to see it in action, especially if you come from the software development side of things where you used to have to code everything. So, I personally use cloud code a lot in my work nowadays because we work with a lot of different code bases and it's a lot easier nowadays with these tools to go into code bases that maybe we haven't touched in two years and find something, find a bug, find an issue, make a feature update because we can very quickly understand and dive into the code. Whereas before, if the code wasn't very well documented, someone else's code, it was very messy and it took a lot of time. Now you can have it run tests. You can do all sorts of things. **[00:30:00]** It's really neat. But getting back to moldbot or open cloth, whatever it's called now, it's very powerful to have this in a chat interface and assume a personality that you give it and continue that thread. So it starts feeling like not so much a tool, but someone something think like Jarvis and Iron Man where where you just ask it what you need and it'll get it done for you. So I think that's where where it really shines. Now again if you just joining us the conversation right now it is important to know be careful when you connect things. Don't expose your APIs to the world. Don't put in a live server that anyone can access. We have it locked tight here on a computer behind the firewall that only we can access. But there's many ways to open it. And and even since it exploded so much in the past few days, you're starting to see these services like digital ocean saying, "Hey, now you could just run a moldbot instance or a open claw instance and just here you go. Like one click, you're done. You can have it on our service and even cloudflare released a version that you can use." So because they're seeing the power in it and they're seeing the interest as well. Like I said at the beginning, I don't know that this will be that thing that people use. But talking about just being honest, it is a rapper in a sense, but so were so many other tools. When Chad GPT came out, hundreds and thousands of rappers came out that solved very specific issues and that was okay. And they made millions of dollars. Some of them, some of them thousands, some of them hundreds, some **[00:32:00]** of them nothing. there where a lot of legitimate applications there still are coming out that are very niche specific and all they do is create a different interface to talk with JPT but focused and oriented on a specific set of instructions that help that particular niche whether it be HR personal development I don't know troubleshooting etc the other day yeah it's a rapper but does it make it way more helpful when it's niched down to something that people actually need or have a way to attract where it's not generalized but but focused on their task at hand. Yeah, of course. And that's what makes those rappers valuable. So, when someone comes out on YouTube and says, "Oh, no, Cloudbot is just a fat. It's just everyone's just talking about it because of the crypto pump and dump or whatever, I disagree. I think it is an amazing tool if you know what to use it for. So, if you have a use case for it, it is an incredible tool. There will come others that do something similar, but this idea of having a way to interact with your different LLMs, that brings it down to earth for a lot of people where you could just interact with it and have it do a lot of things and connect to your accounts and do things just by talking to it. That gets it more aligned to what I think the future of these LLMs is going to be for real people and business owners and everyone else. It's no longer I have to go into this weird interface and just chat with it. No, I I have this assistant that can guide me through things, but then can also fix things and then can also **[00:34:00]** do things and I can also assign a task to do during the day every day and it'll just go and do them. It could literally here's something it could do just just thinking off the top of my head. Imagine you build landing pages not you're not going to do this but maybe someone will. Maybe you build landing pages for a very specific industry like for plumbers or something and you always do the exact same landing page. you're you're one of those businesses that are like, I'll build your landing page with a optin form with uh whatever connected to your calendar and it's $500 something like that. One of those like you just copy paste a template, you just change their name, change their logo, put an image of them and change their services and you charge them like a very basic fee and then you do the next one and you could do 100 per month, right? There are people who do that. This is just very basic example. Now imagine you're doing that. So basically you're going in and maybe you have someone to help you have VA or something and people purchase whatever you're selling and then you go in and you make the adjustments very quickly but it still takes you maybe 10 15 minutes and then you publish it and you're done. Now imagine you can just tell the thing hey we just on boarded XYZ plumbing or Bill's plumbing this is their website go ahead and create it and so or this is their Google places listing go ahead and create it. And so the thing without coding anything beforehand, the thing once you explain it what it should do because it learns it can go grab the information it **[00:36:00]** needs, phone number, services, etc. Go get your template, generate, duplicate the repository, create the little site, go and publish it on your host, connect the DNS, do everything via MCP and then it's ready. You didn't have to do anything. So you can just have it like every time you receive an entry that someone purchased via Stripe, it goes into a Google form or whatever and just do all that things, generate the sites for them. And so 24/7 this thing could be building your landing pages for your customers because you're following that that sort of template. So it's a very basic example of something you could do when you have something that's sort of templatized and how it could go and and do things. And it doesn't even have to be templatized. That's the crazy thing. You can have it where it actually builds unique designs for these people based on their industry, based on what they're offering, based on their location, based on the reviews that they have, based on the their YouTube channel. It's it's incredible. People are using things like cloud code to dramatically increase sales for go-to market teams because now they can use tools like these, not this specifically, but even cloud code or codeex or whatever you're using because they can tell it, hey, I've got this list of people. Go and research them. And I have one of those running and just every day it's just going through a list of people and visiting their websites, visiting the socials, visiting what they're doing business-wise, of course, and looking at interesting data about them and then generating outreach emails or messages or calling them directly with a voice agent. So all that has been possible the past few weeks, few **[00:38:00]** months. So imagine just telling your assistant this thing to do those things. So, you already explain how that works and then you could just say, "Just met this person at this conference. Please send out our package." And it knows to with that send out a an introductory email, send them a tweet, follow them on all those socials, send them a text message the day after how they're doing, send them some tips, and then send them a gift basket or whatever. You can do whatever you want. So, very powerful stuff at the end of the day. And that's why I like it. Yes, it's a rapper, but it's a rapper that was wellmade. welldesigned. It's got a little UI interface that you can go in and see what it uh what it's doing. You can talk to it via multiple channels. You don't have to really set up anything. And that's why it's really nice. So, if you're careful and you're safe and you do it in a machine that doesn't have all your files and private information, then it could become a very nifty little tool that you can use in your day-to-day lives on your business and it will really help you get things going at scale. Really powerful. Now, what are we using it for right now? So, the latest is we're building an application. Two people, different businesses, we're working on an application. So, instead of going back and forth and drafting designs and things like that, we created a group in Telegram and we added the agent, the open claw to it. And now we can both talk to it. And so, we started just brainstorming and say, well, this is what we want to do. Here's the transcript of our **[00:40:00]** previous call where we're talking about it. And then the agent's like, "Oh, sure. Yeah, I understand. You want to do this, this, this, this." Okay, perfect. And then we can ask it to interview us, ask us specific questions to get a more tailored idea, and then just interact with it. And so it's a nice three-way conversation, right? It could be four or five way conversation. And it understands when you're talking to it directly, just like a regular person would in a group chat. And it it's very interesting to see it progress because we can then one of us was talking to it about database schema and structures and how everything should be set up. And then maybe I was talking about UI and design and how it should do that and create some mockups and it's all happening in the conversation and behind the scenes it's generating the schema. It's generating the mockups. It's generating these other things. So our brainstorming conversation becomes actually building the thing and then I can say oh the mockups look nice. Now go ahead and create this in or next or whatever you want. And it goes and creates it. And now well I have this database on superbase. why don't you connect to it and set it up and so it goes it generates the schema that it had goes and creates it by the way you want to do some ingestion data ingestions flows from NA10 I don't want to build them so why don't you build them and so it goes uses MCP goes and creates those workflows in NA10 connects them to superbase for the ingestion grabs RSS feeds does a bunch of stuff all within our conversation and then we already have data in **[00:42:00]** the database and then we say great now let's expand it let's add users okay so it's adds authentication creates the signup form etc. Awesome. Now, let's look at the dashboard. It creates the dashboard. Okay, it looks good. Now, what do you think it's missing? And then it can go in and see like, oh, well, it would be nice we had some charts and data and things like that. So, it goes and self packages for charts, creates the interface, and then says, okay, here it is. Look at the data. And then again, it's it's incredible because it's doing all that work as we only talk about it with the thing. It keeps track and it could be through multiple days and it just goes through and makes it better and makes adjustments and adds things. And since it has access to the data, we can ask it to make assumptions on the data or find trends in the data and things like that that it can do without us having to pipe into a business intelligence tool yet. Like at the beginning, you can have it sort of understand data, parse data, and then say, "Okay, now we have that picture. Where should we put it?" Oh, maybe we can put it in fabric or maybe we can put it over here. then becomes a lot more interesting because you know what to expect and then you can start putting in your different tooling that you might have. That's it. That's what I wanted to talk about. I think it's a great tool. Yes, there was this misunderstanding where it become it hyped up because when Claudebutt started becoming so popular, Anthropic sent a cease and desist type letter to the project's owner and he had **[00:44:00]** to change the name. So in that haste to change the name, some of those handles, the previous one got taken by people who do that sort of stuff and then they started like a memecoin or something. I don't know the whole details, but basically they did something like that. So So a lot of that came out and money may or may not have been lost or generated or lost for many people that that sort of fell into that, but it had nothing to do with the actual project. Just people that that sat on those domains and those handles. So yeah, not very nice, but it has nothing to do with the actual project. The project itself is a very valid interpretation of what an LLM interaction should look like. That's my personal opinion. I think it's great that he created it and that there's a community building around it. I expect more of these types of things to come along and provide better safeguards and things like that for people who who are just jumping on the trend. It looks fun, but not have them lose their data or accidentally share information that that shouldn't be shared. That's a that's a very important thing to consider as well. But if you are in the space and technically inclined even a little bit, then yeah, it's a fun little project that you can do on a separate computer that you may have and and see what it does. I think it's a very enlightening experience to interact with a tool that can do this and I invite other people to try it as well. We've been having a lot of fun with it. It's not perfect. It still makes some mistakes, especially sometimes it lo loses **[00:46:00]** context a little bit and you have to sort of guide it back. But overall, the experience over a few days working on a project, a three-way project with it, it has been very interesting and it has done very powerful things that would have taken us a long time and multiple developers to do. So yeah, I mean if you have a development company, this doesn't mean get rid of your developers. No, you need people to wield the thing, right? So yeah, no, keep your developers, but give them superpowers by giving them access to tools like these that then they work like five developers each and then that gives you more money to get more developers. Then they can work like five developers each, right? Anyway, hope you enjoyed this broadcast and I'll see you in the next one. Make sure you follow and subscribe if you like the content. Right now we're streaming live on YouTube, but you'll find this also on the Web Talk Show podcast on Spotify, iTunes podcast or Apple Podcast or everywhere else that you listen to your podcasts. and I'll see you in the next one.