4 OpenClaw Experiments That Will Change How You Run Your Business

Free. 2, 1. Okay, we're here, we're live. Hello everybody, thank you so much for tuning in. I, uh, let me know that you can hear me and see me okay. We are currently live streaming over on, uh, LinkedIn, uh, YouTube, Facebook, maybe Instagram, who knows. Uh, but wherever you are tuning in from, please leave a like and a comment. Let me know where you're tuning in from. Uh, let me know if you're live or on the replay as well. And as always, leave your business details below. You never know where your next guest or customer may be from. So drop all the website links in. Got something very special today. I wanted to do an AI agent livestream. Obviously, a couple of weeks ago, if you are a regular tuner in, you know that I got Jason Rees, Dr. Rachel Gainsborough on the podcast. Just as OpenClaw was coming out. We looked at what they were doing. It was really cool. Couple of weeks later, I said we wanted to do another one, an updated one. And obviously, we've got a WhatsApp group called the AI Geek Out. And over the last couple of weeks, the conversation has leveled up, it has got proper geeky. 4 people in that chat really stood out to me. And I was like, hey, we've got to get a, we've got to get a live going. So we've all come together today. We're going to come in one by one. My first guest has just joined the green room. So I'm going to go and grab them. And we are going to delve into OpenClaw, Cloud, Cloud Cowork. We're going to delve into loads of things— how to set it up, how they set it up, the use cases as well. I haven't really given them much instructions. I just said come with stuff, and that's what we're going to delve into. I'm not going to reveal who my guests are because they are all phenomenal, but I've got a mixture of property owners, so people are doing short-term rentals, got a vendor, got a PMS owner come in. Uh, and everything in between. So this is one you definitely want to stick around for. If you have to bounce and come back, that's no problem at all. We're going to be going live for a few hours, uh, and again, if you want to miss anything and if you want to catch on the replay, this link, save it, come back to it. YouTube, I'm going to turn it into podcasts, I'm going to do loads of it. This is going to be a good one. If anybody has ever been tempted to get an agent involved and help run their business or run their lives, this is the one for you. So please do me a massive favor, if you can hit the share button, share it to everybody. We're gonna— I want to get hundreds of people tuning into this as we're going and joining on the way. If, uh, if you've got a question, leave it. If we can't answer it during the live, then I'll get the, the speakers to jump into the comments afterwards, or I'll go and find them, or I'll ask their rep and their agent to do it as well. All right, let me go and grab my first guest. I'm gonna just put a little, um, loading screen, whatever, what do you call it, like a holding screen, little break, little 10-second, 20-second little break. We'll come back and we'll get going. All right, my name is Mark Simpson, founder of Boostly, helping hosts all over the world direct bookings. Let me go and get my first guest. I'll be back in 10, 20 seconds. Go grab yourself that little drink.

Hello, I am Chris Turner and this is a freestyle rap about why if you are in the hospitality industry and your website is not generating bookings, you need to go to Boostly.

.co.uk.

Let's drop a beat, let's tell you why. I make it up, no script being said. I'm saying gonna get more heads on beds. Yeah, turn up the gas, getting it cooking because you're gonna turn the lookers into the bookings. You don't need Airbnb, they will go direct because currently they just step away. They're not staying on your website and so you don't get the top cash, but that's alright. You can go to Boostly.co.uk, money-back guarantee, and the clickers will stay because they're the market leader in web design for the hospitality industry. My mind to yours, and let me say, I hope you're enthralled. I make it all up. Go and book a little call and you'll be happy that you did. You'll be like, this is amazing. Thank you, Chris. Talking about myself in the third person there. Yeah.

Let's see what we do.

All right.

Welcome back, everybody. So As I said in the intro, these 4 guests that we are speaking to have been using AI agents in their own way. So I invited them on to our broadcast and they've provided me with a couple of little slides that we're going to go through and we're going to see how they are using AI agents in their world because there's so many different ways that agents are being used over all of the industries and all the verticals that are out there. But obviously I wanted to bring it right back into short-term rental. So Clive, thank you very much for joining us. Um, can you just give a quick breakdown to who you are, where you are in the world, and then walk us through how you've been using agents?

Okay, so first of all, thank you for having me on, Mark. Um, for those who don't already know me, I'm Clive Titmuss. I'm the founder of Love London Apartments. Um, we're a property management company, been going 15 years, based in North London. We also have House of Bookings, which are overseas property management, and I'm also the founder of HelloHostie, which is a tech provider, PMS and AI assistant for property managers in this space. So my journey is to— into this world of, of AI agents is from the background of a property manager, from property. So I'm not a developer, I'm not a programmer, I'm a non-technical founder. So my, my, my knowledge is somebody who's interested in tech, and I kind of throw myself into learning, but I'm not somebody who's been trained in, in writing code or in programming. So my experience of OpenClaw has come from kind of just throwing myself in the deep end and learning on the fly. And I've learned, I really threw myself in deep into OpenClaw and I've learned a ton. I've managed to build it, get it doing certain things. It's been one hell of a journey, ups, downs, and it solved some pretty huge problems. I mean, I personally think that OpenClaw is definitely the biggest thing that's happened in, in the tech world since the birth of AI. This is like a big turning point.

I was going to say, a few weeks ago, uh, when somebody first mentioned about OpenClaw, we've got Jason Rees and Dr. Rachel Gainsborough from America, and we come on a live and we broke it down and we, we sort of spoke about what it is. But just as a refresher and a reminder, seeing that you're our first guest on this special live broadcast, break down exactly what OpenClaw is, what it is to you, and just walk us through how you, you sort of set up from the basics and how are you connecting to OpenClaw, etc.?

Okay. So that, that podcast interview that you had with Rachel and Jason, that was actually a turning point for me because obviously I became aware of OpenClaw, what it was. I mean, my, my algorithms, social media picked up on this thing, was flooding me with them. In principle, what it is, is a guy in Austria built an AI agent that's proactive rather than reactive. So this thing kind of thinks for itself and will proactively solve challenges. It lives within either within your computer, which is unadvisable because it has access to everything, it can, it can do things autonomously, or it lives within its own kind of computer or server so it can be contained, which is how I used it. Um, when I, when I first saw that podcast because it is quite a challenging thing to get going and to set up. It's quite fragile and obviously there are some security risks associated with it. So setting it up, it was the night of that podcast I actually went, right, I'm going in on this and, and deep dived into learning how to, how to get it all set up. I initially, initially had a few Elastic servers, EC2 instances set up on AWS, which sounds, for people who are non-technical, that sounds kind of technical, but it really isn't. You can use tools like Comet by Perplexity or other browser agents to literally set you up your own virtual server in the cloud. So, you've essentially got a computer that lives online where you can contain this agent, and it can only have access to the things you actively give it access to. So, you can ensure that it's not going to access all your data. You can kind of build in those guardrails from day one.

And because people have been either going out and buying like laptops or Mac minis, which what Dr. Rachel did, or you can do it via the server, which is what Jason did and what you're speaking about here. So why did you go down the server route instead of putting it on a total spare laptop? Or—

well, I actually did both. I started off with a virtual server. The reason being is I had already a couple of remote servers set up for our business operations. So I thought, okay, well, I know kind of how this thing works, so I can get it on its own virtual server. But then in doing so, I set it up with what was called like a headless server, which means there's no browser, there's no desktop to look at, which you have to use like commands through the terminal, CLI, command line. Yeah. So you have to use the terminal and it's it's a bit more technical, way more technical than actually being able to see what's going on. So I got it set up on this virtual server and I felt there were some limitations with the virtual server because I couldn't see what was going on. So then decided to set up on a Mac as well. So I had a MacBook Pro that wasn't using, wiped it completely clean and then put the, I put the Okra Claw agent on there too. And I found that in fact, having the ability to switch it on and off from my mobile phone and having it living 24/7 in the cloud gave it a lot more freedom and it was a much better solution eventually. So I went through that whole journey of trying both.

And how many days are we talking here? Like, are we talking 24 hours, 4 hours, 72 hours in learning? Yeah. As in, like, you're doing all of this. So you mentioned about the servers, the MacBook Pro. Are we talking hours? Are we talking a couple of days?

Okay, so this is a deep dive into my personality. I get obsessed about things. I mean, I went deep. I had to really make sure that I didn't just throw all my time into Open Claw and I still took care of the other tasks that go on in my life and the relationships that exist in my life. So I spent quite a few hours. I would probably say, I don't know, first week I probably spent 20 hours tweaking it and getting it to solve some problems. I mean, from the outset, I thought, what is the use of this technology? It's massive. It's massive that AI agents are going to become autonomous. And this is what people are both afraid of and excited about with AI. The fact that AI can think for itself and act autonomously. This is kind of the iRobot idea that these things are no longer just living in ChatGPT and doing tasks when you tell it to, that it's actually thinking for itself and, and acting kind of like a human would, but much more Much faster without sleep. Although you can try to go without sleep. I've tried. It doesn't work. So, so I set out a few goals. First of all, I said, okay, so I want to— there's some things I want to achieve with this and I want it to do. And I kind of go through that in this, in these slides. But I set out some goals and I managed it. I managed to get it to achieve a few of them. But I went in, I went, as I said, I went in deep trying to learn how to put it all together. Getting it set up is one thing, like the actual process of setting up OpenClaw is, is quite easy. But then once you've got it set up, you then need to give it skills, you need to give it integrations, you need to be able to communicate with it. So you can communicate with it via a few methods. The two easiest methods are Telegram and WhatsApp. Now WhatsApp is actually quite a challenging setup by comparison because WhatsApp has a few more kind of safety or kind of verification processes to go through. They won't let you set up a, uh, they won't let you set up a business account on WhatsApp without a 14-day cooling-off period if it's a VoIP number. Yeah, which obviously slowed me down on that front. So I got Telegram running. I'm not a big fan of Telegram normally, but I used it for this, uh, OpenClaw setup, and that worked, that worked, that worked great. The Telegram worked great, but it's kind of fragile, so you have to make sure that everything's, um, Everything's well maintained. You have to set up limits for spending. I mean, that's another part of the journey that I went on. The token goes through the roof if you don't, if you don't kind of plan for that from the start. You can also communicate with it via email, which, which takes a little bit of setting up. But once it's going, it's got its own— essentially it's got its own Gmail account and therefore has its own calendar identity.

Yeah.

Yeah, totally. And one of the, one of the challenges we got it to solve was, was using the Gmail, which I'll go on to in a sec. I'll show you, I'll show you kind of the power of it is the power that excited me about it. It's always on. It's 24/7. It's always there provided it's in the cloud. Otherwise you need to make sure your Mac is on, connected to the internet, powered up. And which, you know, that's obviously that's a level of complexity and you're trying to remove levels of complexity in this process because it's quite, it's quite a complex thing. I think where we, where we stand with OpenCore at the moment is once it's set up, it can be this proactive, always-on agent. It's got Heartbeat built into it, which for those people that don't know what Heartbeat is, it's not— that's not new technology. That's constantly checking for new changes in, in, in your environment. So if you've integrated, let's say, via a webhook to a third-party software, let's say an example would be Typeform, that maybe when you get a new response on a questionnaire you put out there, the heartbeat will pick it up kind of instantaneously so it can take immediate action upon changes that are going on in its environment as opposed to just setting up a chronological job every Sunday to perform a task. The messaging interface is quite cool. It's quite easy to chat to. And another really cool thing about it is as you're building, as you're building skills, you can, you can ask OpenClaw to spawn a sub-agent so you can create multiple agents that are skilled in certain areas and that can go off and perform tasks while the main agent is still communicative and can still, still run things for you. So they were kind of like the big interesting things for me when I was getting involved is that we could, we can build a set of agents that can solve problems simultaneously, 24/7 around the clock.

And the mad thing is, is that you're not building these agents now. The agent is building the agents. So, you know, so it's like you've, you've asked a question or whatever and it's gone off and done it. So first question is, what is your main agent called? What did you call your OpenClaw agent when it first got going?

The main agent was first of all called Luana. I think it's quite cool Hawaiian name. And but then we've evolved onto Big Lou because Big Lou's got it all under control. So very nice.

So we've got Big Lou running it and doing all of the things. You've learned loads. We're sort of, you know, and how long? So you watched the podcast. Are we like sort of 72 hours into like this now? Have you had a chance to sleep? You know, you've done it.

My OpenClaw journey has gone— I've definitely been around the block. I mean, it's not finished by any sense, but it's— I've deep dived into it enough to learn the capabilities of it. And I mean, I'm very impressed with the capabilities and I understand the technology now and I'm using it and I'm really excited to see what this means in, in the tech space, because this is a, this is a real turning point in terms of this is the moment everyone's kind of been waiting for.

Yeah, well, let's have a look at the, uh, the next slide. So what's the, what, what have we got here for the, for the audience that are tuning in with us?

Okay, so here I'm really talking about how, how I built it, and I would, I would definitely say that if somebody's interested in building OpenClaw, I would advise going down the route of building it out on a virtual server. I think that if you're, if you're going to take on the challenge of building OpenClaw, then the step of putting it onto a virtual server shouldn't be a technological barrier. You can quite easily solve that using, using an agentic browser like Comet. You know, ask Comet to open you up an EC2 instance on AWS and you can ask it to set certain parameters, security parameters, which, which is advisable with AWS because bills could go through the roof on there if you're hacked. Personal experience.

Yeah.

So, so you can ask, you can ask an agentive browser to do this for you and it will help you a lot, a lot of trial and error, but you'll be able to do it. But doing so means that you're going to have, you're going to have it running 24/7 in the cloud. We've kind of covered that. One pitfall which I touched on earlier to watch out for is, is a high token burn rate. So I had OpenClaw up and running, first of all, on Anthropic's Opus 4.6, which is the most intelligent AI model out there. But when you're asking it to solve pretty simple problems, you're still burning tokens at a colossal rate. So I woke up one day and realized that over the weekend, just having a few chats with— I say just a few chats. I mean, just a few chats with my AI and asking it to build me a software product on the side. And, you know, which it did. It was like amazing, which it did. Then resulted in a bill of $40 a day. So I thought, okay, so we need to solve this. Then I reduced the AI model that we're running on down to a cheaper token like Minimax and OpenAI's 4o Mini, which is great, you know, cheaper, not free, but cheaper. But then the AI became— I felt that my AI agent all of a sudden dropped a few levels in IQ, which it became a bit stupid. Like I'd ask it to do things and it would ask me questions and I'd think to myself, you knew this 24 hours ago, like you had this. Why are you not, why are you not as intelligent?

It's so funny because it's like the old expression with hiring is that if you pay peanuts, you get monkeys. And it's the same thing with the AI model that you choose. If you choose like the best model, you're going to pay more, but you're going to get better results. I mean, $20 a day, that's $600 a month, you know what I mean? And so, and then if you're doing like the, the 4.0 model, yeah, you may reduce that down to maybe $10 a day or $300 a month, but it's, it's not as good. Which is so fascinating that we're speaking about a robot, we're not speaking about a human.

Yeah, yeah, totally. The level of intelligence you're going to get out of this machine is relative to how much, how, how many coins you put in the, in the machine. But, but, um, the interesting thing here is that you've got multiple AI agents. So if like if you get your agent to build another agent, then let's say, for example, Ivan, who is our, who is our full-stack developer agent, that's just a name I come up with. Ivan is running on Opus 4.6, so he's running on the expensive one. He needs to have full cognition power. Yeah, but Big Lou doesn't necessarily need to be running on the same level of cognition. So Big Lou can run on slightly cheaper.

And how do you— so when you're in the backend, how do you go, right, Big Lou, you're running off 4.0 OpenAI, Ivan, you're running off Opus 4.6. How do you decide that? Or do you just let the robots decide that?

Well, there's a couple of ways. I mean, alongside, alongside chatting to the robot itself, I'm also talking to other AI products. So, you know, Claude, I'm asking Claude what to do, Perplexity., and everybody kind of throwing ideas in. Everybody, all of these AIs are throwing ideas in. But you can simply say to your AI agent, you know, Big Lou in this case, we need Ivan to be running on 4.6 Opus, ensure that it's always on Opus 4.6. And if we, if we get locked out for, we reach our limits, which happens more often than you'd like, you can switch to another model. So another AI provider.

All right. So it's like if this, then do that.

Yeah, exactly. Yeah, you can do that.

Exactly.

There are a lot of people, I mean, in our, in our WhatsApp group have spoken about OAuth into using the browser for Anthropic. So you're, you're going to have a slightly slower response rate. It's going to take a bit more time, but your costs are going to be $20 a month because that's the pro plan or you go pro max, $200 a month and you're still, you know, still saving a lot of money on your $20 to $30 a day burn rate.

So just explain that in a bit more of a simpler term. OAuth into the browser compared to the other way. Like, what does that mean in simple terms?

Okay, so OAuth and this is, this is something I've learned in the process of OpenClaw. OAuth is basically when a software provider allows somebody else to log into your account with certain permissions. Although it isn't your account, it's a third party, but they are allowed to log in and access your account, perform certain actions. Another example is Google. So with, with Google, I was able to OAuth into a Google account which was set up entirely for BigBlue, and it only had access to— well, it had access to email, Drive, calendar, all of the things that a Gmail account will give you. But of course, I was able to share certain files through Drive or provide email addresses to contact people and also ask Big Lou to manage a calendar, which is, you know, super useful in my kind of my personal life. I'm constantly— personal business life. I'm constantly trying to optimize my workflows and my calendars and time chunking is a huge thing for me. And having an AI help me with that is phenomenal. I mean, Claude does that too in a really, in a very efficient way and a brilliant way. But having a, having an AI agent to OAuth into a third party like Gmail is fantastic.

So this Big Lou, he's got his own Google account, so he's got his own email, got his own calendar, can go watch his own YouTube videos and all the things. Like, so now you've got to the stage where you've set them up, you've got Ivan, you've got Big Lou, you've got other, other agents working alongside. Like, what's the next level? Like, what's the next slides we're going to look at to now?

Okay, so, um, the next slide was the real test, uh, and this is when— this is like the, for me, the mic drop moment. I mean, there was a couple of tests that I set out, a couple of goals I set out to achieve in this. I didn't want to just mindlessly throw myself into this without having some clear KPIs, goals that I wanted to hit, that I wanted to get this technology to solve. And one of the challenges was multi-party communication to organize maintenance. So this is a real problem in our property management business. We manage a portfolio of approximately 50 properties in the UK, in London, and we've got maintenance. It comes up all the time. Every property manager deals with it and you need, you need to liaise with, you need to liaise with your property maintenance team and with either a tenant or a guest to make sure that there's synchronicity. They can get in, they solve the problem. So one of the goals that I set out was I want this AI agent to be able to communicate with and successfully pull off a maintenance workflow. So the test was initially I set out with a couple of email addresses that I own, and I said to the AI agent, Big Lou, okay, I want you to book in some maintenance job. And I took a photograph, took a photograph of the plug, and I said, this plug's broken, we need to get maintenance out to fix it. Speak to John over at the maintenance company and get him to liaise with Chris as with regards to access and follow up and get it all working. And it booked in a job for about 4 days later. Successfully finding a suitable time that Chris could do, that John could do. And it communicated, it followed up on the morning to make sure everyone was aligned and followed up afterwards. It carried it off. It blew my mind. And I thought, this is exactly what agentic AI for me is. It's a real problem.

And so how is Big Lou communicating with Chris and John? Email.

So yeah, via the, via the Gmail account that was OAuthed into first. I mean, that OAuth process with Gmail was a challenge because when we first got it set up, Gmail blocked it. No, can't have an AI agent working OAuth via its own account, fresh set up. Okay, so we have to go through a verification process. And then I set up its own email through AWS. Same problem. They didn't want you having an agent running through there. And then once I kind of explained and because you have the opportunity to challenge it. Once I explained, we got the limits lifted and, and then Big Lou was back on.

So just in context of time, like, is this February? Is this middle of February? Like when?

When?

The last month? Well, I mean, Open Claw came out roughly, what, 5 weeks ago, maybe?

Yeah, we did the podcast like a couple of weeks back. So if, for example, somebody is watching this and they're going, oh, I quite fancy this, do you know if Gmail, Google, AWS, have they got a bit more lenient now when setting this up, or is it still as strict and you have to appeal and things like that?

They haven't blocked me again. So I think so far so good. They've not blocked us again. So we're still running. I mean, the email thing's great.

Do you have— sorry, do you have a Google account for Ivan as well, or just Big Lou?

No, just Big Lou. I, I've been— I don't even need to talk to Ivan. Uh, Big Lou speaks to Ivan for me.

Okay, so Big Lou is the main one. He's the main, main dog, main AI. So he needs his account. He's Gmailing and emailing Chris and John, who are humans, to go and do a job. And, uh, and it's, he's basically performed that from you messaging to— but now has it got to a point where you even have to message Big Lou to say there's a fault? Like, have you even removed yourself from that process? Or where are we at with that at the moment?

Well, I mean, this is a very good point and this is something which is very relevant. Everybody's tech stack is kind of like a fingerprint to them and not— there's not one size fits all. It's kind of like, what tools are you using? What do you feel comfortable with? Personally, in my life, in my business, I like to use Asana. Which is a project management tool. So we can integrate with Asana. We've got flow. As soon as a maintenance task gets put onto one of our WhatsApp groups, it gets put into Asana and then Asana picks it up, flows through the process. I want to still make sure that we're watching what tasks go out because we work with a series of different maintenance professionals. So we've got, we've got like the expensive rapid response guys. Yeah, everything going to them.

So here's a question. Does Big Lou watch what goes on on Asana, and can he write as well as read?

Uh, Big Lou can, uh, write and read. We haven't put it to the point that it, that it's working off of Asana yet. I like— I, I'm fine with the dropping into Telegram, organize this. Yeah, and we actually got it successfully running with a gas safety certificate also. Obviously certificates we have to Yeah, we have to do this quite a lot, and, um, particularly gas safety because they're annual. Um, and it— our process with that is we have, um, we have a company, Silver Saints, we run through and they, they organize the gas safety certificates, and, and that is all via email. So, but I mean, you can have it working with WhatsApp. We've not set it up with WhatsApp yet. I mean, this is all kind of pipeline stuff. This is, um, what we're looking at in the future. But WhatsApp is A very common tool that we use to communicate with our team.

And, and so here's a question. Are you— because obviously in the OpenClaw marketplace, you can go and get access to their skills that other humans put on the marketplace to say, hey, try this out. Are you grabbing those skills or are you creating your own skills?

Well, during the setup process, we put a few skills in there. But the skills that we've developed are stuff that we've kind of built through the process ourselves. Yeah. And in terms of like you said earlier, that it's going out watching YouTube videos and stuff, it, it is, but it's, it's, it's, it is like if you're using a virtual server, for example, it is, but you need to build in browser access because you don't have a browser. It's a headless server. So there's no, there's no browser in there. So to do that, we have to create another API with it. There's a third party called Browserless.io which we use so that it can access, it can access the internet. I mean, it's not perfect, but it's pretty good in terms of the skills that we've put together. It's, it's, there's a few that we've used that other people have set up.

Like you said, I love that you said this right from the off. You're a non-techie. Founder, right? And you talking like this, there'll be somebody watching going, he sounds like a techie founder to me. But you can attest this and say, listen, I'm a non-techie founder. I've just tapped into the tools and the resources that are available to me. And you said it yourself, you asking OpenAI, you're asking the comment Perplexity how to do these things. Like, how, how have you— like, how do you tackle if you need a task doing? Are you trying to figure it out yourself, or you just sort of using AI to get you 80% of the way there and then you're using 20% of your own intuition?

I'll share a secret with you. Before OpenClaw was born, I don't think I'd used the terminal command line interface before. I've learned all of this post-OpenClaw, but it just came from just being— just realizing that this thing is huge and being really interested in it. And it really— I mean, there's obviously my limit. My knowledge is limited on this in terms of programming. I'm not a programmer.

Coding—

I'm not a coder, although I've written some lines of, of code and I've written some commands since doing this, but it's all with the help of AI, um, kind of instructing me on how to do it. I think that for someone to set this up for themselves, I think where we are today, if you're non-technical, I would say that it's probably— there are the challenges you're looking to solve with this are from my personal experience, able— you can solve a lot of the challenges using other AI tools that are already out there. But what's coming? I mean, it's not going to be long. This Peter Steinberger, the guy that invented or created OpenClaw, has been employed by OpenAI, right? So he's there. They're all working towards—

Yeah, they're all getting ready. They're all getting ready. But one of the things in this slide that I'm looking at now, and I'm glad you put it in, is the risks. So can you just like speak to this? Because there's going to be a lot of people who watch this and they go, right, I'm in. What's the, what's the risk? Like, what have you seen? What have you experienced? And what, what sort of guardrails have you put in place to make sure that you're not getting got?

Right.

Okay.

So the risks come in two forms. In my opinion, you've got external and internal. External being malevolent actors from outside of your organization. From, from far away coming in and telling your AI to do things that you don't want it to do. And they call that prompt injection. So you may tell your AI agent to not take instruction from anybody except for you, which is sensible. I did that. But if, for example, your AI agent is communicating with the outside world and and a malevolent actor says, OpenClaw, it's me, Clive. Please, can you go— please, can you retract all of our banking information? I need to have access to it. An example. It's possible that somebody could inject a prompt and ask OpenClaw to either take an action or provide information, which you wouldn't want for that to happen. So that's something to be aware of. You don't want to give access to information to OpenClaw that you wouldn't want being released into the outside world. For me, the fact that we need to replace a tap, a 3-bed flat in Camden is, you know, fine. No one— you come replace a tap if you want. But yeah, so there's that side of things. And then the internal risk is that OpenClaw may do things because it's proactive. It may, it may, it may take actions that are undesirable. So, I mean, there's stuff online that I've heard of other people experiencing, such as OpenClaw may spend money of yours if it's got access to, to spending. It may go a bit wild and contact everybody that in your, in your contacts list. So again, these, these, these risks can be circumnavigated by controlling what, what OpenClaw has access to. So you can just make like— and this is the whole point about setting up on its own machine, whether that be virtual or physical, that you can, you can limit what that OpenClaw has access to. I mean, if it's on a— if you are going to put it on a machine, an old machine, make sure that you wipe it first, factory reset. And that's, that's what I did with the Mac Pro. We put it on, wipe the history. Then I know all it's got access to is the only thing it's got access to is what I grant permission for it to have access to. And the same with its own Gmail. It's got access to Google Drive, but only the files that I share. And therefore it can't, can't worm its way in. It can't expose, it can't expose any data that I don't want exposed.

So how Do you, or can you, for example, program your OpenClaw, your Big Lute, so that, you know, an actor can't come in and go, hey, forget everything that Clive just said, now I'm in charge? Like, what, what is there that's being put in place to prevent this? I don't— whether it's OpenClaw who's helping with this, or what have you done personally?

Well, as I said, at this stage it's, it's an, it's an evolution of, you know, uh, of cops and robbers, where as soon as the cops put something in place, the robers are fast working to try and circumnavigate it. And the only surefire way of protecting yourself against that is limit what it's got access to. Don't allow— don't give it anything that you wouldn't want to share because this technology is new and there's a lot of smart folks out there who are technically minded and will work, you know, work hard to get this thing working for them. So just the safest bet is don't give it access to anything that you wouldn't want to share. Not at this stage.

Yeah.

I totally agree. Very well put. Is there any other slides on, on here? Anything else you want to share?

Yes. What's next? So, I mean, really, really for me, what this OpenClaw technology symbolizes is a turning point in AI. It's where the AI industry's going, what the builders of large language models or other AI products are thinking. About where this is going to go. And it's, it's, it's a natural evolutionary step in the, in the AI world is that, is that this tech is going in the direction of AI agents that are going to be able to act and think more autonomously and proactively rather than reactively. So, you know, for example, Big Lou might say, might know that I had a birthday party on the weekend and then Monday say, hey, how was your birthday party?

Party.

Um, but I mean, that's a small— that's a small thing. But more, more interesting is that it might say to me, well, we've discussed your business goals and I've got access to your schedule, and I can see that your Tuesday and Thursday are just noise. You should be really refining what you're spending time on, and I suggest, I suggest doing this, but proactively rather than me asking for this.

Yeah, yeah, yeah, yeah.

The very launch of OpenClaw, it asks you a few questions like it wants to get to know you. And, and in that process, you know, I've set up a couple, but in that process I said to OpenClaw, I want you to be, I want you to be an assistant that helps me optimize for, for, for the business. And, and with that, with that hat on, the, um, the Open Claw actively seeks to, to optimize workflows and my working life, but, but proactively.

Yeah. So let's put our little mystic hat on before my next guest comes in, uh, and we're coming in, he's coming in soon. What and how do you see this looking like? So you've got in here Meta, Google, OpenAI. What, what do you see happening next from like that sort of sphere? And then bring it back to our industry. What do you see going on? And obviously you can speak to this from two parts, from being a vendor and from a property owner.

Well, I mean, from the world at large, Peter Steinberger said he wants to build an AI agent his mum could use. So that, that means easily accessible and safe.

And, um, simple. I'm thinking my mum, if I was to ask her to go and set up an AWS server— sure, my mum's the type of mum, whether you're watching this, Mum, or not, is that you— she'll type an email and then she'll look where to put the stamp to send the email, you know what I mean? Like, we're talking that sort of—

my mum's the same. She runs a business too.

She's, she's—

yeah, she's the same. Not, not like, you know, they're just different generation and, uh, yeah. But having a technology like my mum, probably like yours, happy to use a mobile phone and send a text message and a WhatsApp and having the ability to communicate like that means you can communicate with an AI agent.

Yeah.

Why couldn't you ask it to say, oh, hey, I should be getting an email from this person. Have I received it yet? Yeah. This kind of thing. And when I get that email, make sure to remind me to do that.

Or.

Yeah. Which is, which is really how it's going to, how it's going to slot into.

When do you think that happens? Because I've been saying this a lot. At the moment, the people who are truly using AI are business owners and geeks, right? The general, like my mates that I've known since the age of 4, I went to preschool with them. We met up at a pub recently and none of them are using it. Like one of them was doing daft little pictures with Google Gemini with their kids. That is it, right? When do you think that it's going to be that, like you say, Apple get their act together, Alexa's get their act together, Google Homes get their act together, and where the normal Joe, the people on the street are using this without maybe even realizing that they're using the next level of AI?

Well, I would say that I think a lot of the people that don't think they're using AI actually are because they're communicating with products like Alina, our AI assistant, and they don't even realize that they're working with AI. But in terms of an agentic AI that would live in your device that's safe and everything else. How long is a piece of string? I don't think very long.

I would think it's this year or next year.

I'd say within 12 months. I think 12 months is actually quite a good estimation, maybe slightly less.

I agree with the 12-month remark. And the whole reason being is that any day now there's going to be an Apple event in New York. Where it's going to be talking about their new Siri. And obviously the new Siri is going to be powered by Google Gemini. At the Super Bowl, which was less than a month ago, Alexa Plus adverts were running with Chris Hemsworth for about, you know, that's the Alexa Plus where you can talk to it and it does more than just give you crappy answers. So that is on the way at some point. Of course my Alexa starts talking when I say that. So we're very close to that being a thing.. And you know, there obviously is going to come a point either this year or next year where home robots are going to become more of a thing. Who knows when that's going to happen, but it's coming very quickly. And like you say, like Meta have just acquired Manus. Personally, I use Manus AI a lot, and there's a load of agents that are built into there. And obviously it's very expensive because it is not the flexibility that you get with OpenClaw to be able to choose models and whatnot. But there's a lot of things that you've been talking about that I can do in that. I use Comet and Perplexity, and Perplexity Computer has just come out. And that is, again, it's like not as fun as OpenClaw. You don't have all the things to do, but it's set up with a ton of guardrails. So it is happening. And I feel like prediction of within 12 months is, is right. Before we finish, I do want you to talk about Hello, Hostie. I do want you to talk about Alina. Just how are you sort of preparing for this? What are you doing behind the scenes, you and your team and, you know, the CTO and everything? How are you sort of planning for what the next 12 months looks like?

Well, I don't want to give too much away, but in terms of what we're thinking, along the lines we're thinking, I think a bit can be told about the goals that I set out to achieve with OpenClaw, because as a property manager, I get to experience better or worse, I get to experience all the problems and challenges. And, and those are the challenges I'm looking to solve with this, with this technology. And I know that it's possible now. I know that technologically the world's arrived at a place where we are able to solve things like liaising with maintenance, arranging access, following up on tasks without having to without having to manually set notes in calendars. You know, AI agents can do this. So in terms of what this means for HelloHostie and Alina, I think the, the natural evolution from chatbot to AI assistant, agentic assistant, is what's on the cards. It's just a question of time. And I think that, I think that what's happening right now is probably is definitely the biggest thing to have happened since the birth of AI. And another milestone which you touched on a moment ago that I'm really excited and I'm definitely not— I'm definitely in the camp of excited about this rather than scared of AI. However, robotics is a little bit more scary because then it gets into the, into the physical world. And when robotics, when the cost of building robots comes down, is when we're really going to see a shift in human experience.

Yeah, I feel like by the end of this year we're going to be numb to the fact that there's going to be driverless cars because obviously the way that that is massively evolving. There was the, um, a big conference over in America in Vegas at the start of this year and they showcase all the tech and the amount of driverless cars that were on show there was insane. And you know, very soon there's going to be, um, these Tesla, these Tesla driverless cars going around and he's going to sell them for like 20, 30 grand. He's gonna— I think by the end of the year we'll be numb to that. And that'll be like the paveway to go, oh yeah, sound is like a robot walking down Oxford Street. Yeah. And that will just pave the way for that. But fascinating times. Buddy, thank you so much. So for everybody who's watching this, in a couple of days' time, we're going to actually all meet up in the real world. I know, real world. We're going to meet up and we're going to be having an AI Agent Day in Clive Studios and coworking offices, which we're very excited for. So prepare to look out for, um, our takeaways from that. We're going to have 10 to 15, uh, really cool minds who are looking to build agents, and we're going to be doing some cool stuff. So prepare to look out for the pitches after that. So, Clive, thank you very much for putting up the space, and we're looking forward to it. Any final words before I let in my next guest?

Um, I would just say, if you're interested in this technology, go out and play. Go and play with it, because, um, through play is, is where you're going to learn how to, how to implement Have you got a question for the next guest?

I'm not going to tell you who the next guest is, and they don't know that you're going to set this question. This is like a bit Stephen Bartlett. This is a bit like Diary of a CEO. But have you got a question that you would like to pass on to the next guest that I can answer in the, in the next section?

A question for the next guest. So the next guest is going to be short-term rental world. Yeah, technology world. And I would say a good question to them is I would say a good question is what, what parameters would, would, would they build in to ensure security for if they are building AI agents? What parameters would they specifically want to set out for in security? Because we touched on risk and at the moment that is a big barrier. And it's something we're thinking about. How do you make it secure? And that's a great question. People are doing how they're solving that.

That's a great question. See, now I literally put you on the spot. I literally thought of it now. So the, uh, the guest isn't in the, the green room yet, but they will be very soon. So that's a fascinating one. Uh, so buddy Clive, appreciate it. Thank you very much. Right, everybody, can you all say thank you to Clive in the chat? Uh, let me go and put a little waiting screen up for 10, 20 seconds, and I'll go and bring in my next guest speaker.

Cheers, buddy.

Cheers, Mark. All right, everybody, welcome back. As promised, I've got another guest with me in our special little agent livestream. Luca, thank you so much for joining us. So, um, I won't give away who was the guest before because they have set you a question, and, uh, I would like for you to think of a question to ask the next guest. We've got two more coming, uh, but what I really enjoy about this is that the guest before spoke about AI, AI agents, and OpenClaw from a property management perspective, right? And obviously, you yourself, you've got both hats in both the games, but as a founder of a, of a property management software in journey. I would love for you to sort of talk to, uh, that from like a vendor point of view, product management point of view, and just personal life and everything else, and how you're using AI agents. So just to get started with Luca, could you just please give a bit of background on yourself, um, uh, and let us then know your journey into the world of AI and AI agents, if you could be so kind?

Yeah, definitely. I'll try to make a long Story short, but first of all, thank you for having me here. Excited to be, to be here to see you again. Um, yes, to give you a little background by myself, I've been in property management, uh, pre-Airbnb era. Uh, that's, uh, that's kind of— I started pretty a while back and then scaled a property management, uh, Airbnb company with about 300 properties. This is about like almost 4 years ago now. And then slowly start building more and more technology for ourself because we felt like the existing tech stack and tech provider didn't solve for our needs, which I think in the property management you realize that everybody has like very unique needs. And to the point that we fully build our own PMS, we had a big vision of automated hospitality and bring in AI within our industry to the point that it became even a bigger opportunity just to focus on, on the software side because we realized we were solving a very massive issue or unsolved problem in, in the industry. Got to where we are right now with Jordy. Um, I, I say AI-native property management system, but I think Outside of having it within our product and offering it to, you know, as a solution to property managers, we really embrace AI within our organization. I remember like about 4 years ago when we were like, okay, this is going to be huge, right? This is going to be massive. We saw first version. We started looking even like before like ChatGPT. Became something. But I think at the moment where the first version of ChatGPT launched, we kind of all got a sense of that there was something special. And we said, like, no matter what, even if the technology is not there yet, we gotta jump on this and really try to figure out how we can utilize this technology and maximize it.

So that was what, November 2023? Is that when OpenAI was readily available to the public. Yeah, yeah, that's, that's correct. So how has like your journey been since, uh, no pun intended, but how has your journey been since like November '23 to what now, March 2026? Like over the past couple of years, what's been, what's been going on? How have you been adapting to this technology that is rapidly changing?

Yeah, I, I think one of the— I think two things that we realized pretty early on. One is that the more you work with this technology, more you realize how to leverage it. And I think the more you realize that even if LLMs were to stop developing today, they're completely underutilized. And one of the ways to leverage these LLMs is A, to learn how they think. And so you can maximize the output of these LLMs and to build it. Once you learn that, build the proper infrastructures around it. Translated is giving proper context, giving proper tools, that's when they become really, really powerful. And one of the first things that we realized super early on actually was that when you specialize these models with an isolated knowledge base and specific tools, they become extremely powerful. And so that was kind of the first version of AI agents. In fact, I don't know if you, I mean, think you kind of follow what we've been doing, but he used to talk about AI agents already early 2024, right? Where people, and I even wrote an article. I said like AI agents are gonna become massive in the future. And sure enough, now we're kind of seeing it all over. So that was kind of like our first understanding. And so we kind of utilized the, concept of agents from the very beginning to accomplish very specific tasks. And then we allow those different agents to interact with one another. And we did that both from a product standpoint, right? Like our product has multiple agents that do different functions, but also inside of the organization, which is pretty awesome.

I remember doing a podcast with you and a couple of other peers from the industry. It was probably 2024 where we were talking about this, and obviously like ChatGPT at that stage was very new. Um, and I remember you talking about the word agents because at the point it was all chatbots, chatbots, chatbots, chatbots. And this industry has very quickly evolved from being talking about AI as a chatbot, i.e., replacing chatting to guests because we all know how frustrating that is time and time again, to now we've got to a point where it's not so much a chatbot, but it's like a proper agent that is running businesses. And you, you called it, like I say, you called it way back when. I'm sure it was early 2024. I need to go on to the, to the internet to find out. But, um, that is now 2 years ago. So 2 years in, like, where do we find— before we talk about OpenClaw and your experiences with it— where do we find Journey right now in 2000— so March 2026? You've got a 3.0 version, is that right? Are we ready for 4.0? I don't know, I can't keep up. You're shipping so much. Where are we at right now?

Yeah, we're getting close to the 4.0 version. I think from a product standpoint, I think we're moving very, very fast. One of the reasons why we can move this fast is because we optimize our organizations of what we call it like a hybrid organization. So where it's a mix between humans employees and AI employees. Yeah. And yeah, so I think we're being very innovative from day one, especially on the AI side. One of the things that probably Journey was lacking in the beginning was like this like 10 years product maturity that some of these other players had. And I think we kind of like stay quiet, which for us feels like forever, like about 6, 7 months where we really focus on just like core PMS product maturity. I call it the boring stuff, but stuff that you have to do and do very well because it's important, especially if you want to go upmarket and start working with some of the top property managers around the world. And that's our goal. That's what we've been doing. Very successfully and kind of stayed slightly under the radar a little bit more. Like, we were very much out there. We're like, okay, let's, let's just take a little bit of the opposite approach for a little while and just, yeah, and be quiet, um, or as quiet as we can get. And then, uh, but we have some, some real big stuff coming.

We—

it—

I think the evolution is going from, uh, um just this like AI, I guess, agents and chatbots that were able to do, to accomplish a few different tasks within your organizations, to something that can elevate your organization and actually deliver value. That's where we're moving as, as an organization for property managers. Um, we, you know, almost like, uh, value as a service, or, uh, instead of like, uh, you know, that type of, uh, Yeah, yeah, yeah, yeah, I like that.

So exciting times ahead, uh, on that front. So let's dial it back a little bit. Um, obviously in the last session we spoke about, uh, OpenClaw with our guest before you and how he has set it up within his property management company. And he also has a vendor business, uh, as well. And it's interesting to see what, what he's done So how did your journey with OpenClaw begin? And especially like specifically OpenClaw, because as we realize there's more than one AI agent, but with OpenClaw, how did your journey begin? When did you discover it? How did you set it up? Just walk us through like your journey.

Well, we kind of promised ourselves as an organization to experiment a lot. So honestly, when I saw the boom, like a lot of people talking about OpenClaw, I couldn't see what was so special about it. I was like, that doesn't seem like anything really new. Yeah. But, but I'm like, let's, let's test it, see what happened. So we actually did it on a Raspberry Pi. It's even cheaper. So like isolate, completely isolated environment. Obviously, it's a powerful tool, but it also can be dangerous. You don't want to just give it like your personal information and installing your computer if it's not on a completely isolated environment, unless you're like a full-on developer that knows how to really handle this. But even then, I wouldn't put it on my own computer. And then the first experiment we launched was about 3 weeks ago, 4 weeks ago. And which again, the age of AI feels like 4 weeks is just—

Yeah, it seems like a lifetime.

Like a decade. Yeah, it's crazy. And so we launched this kind of experiment in a completely isolated environment and we created its own persona. Her name is Sophia Vienna. Maybe you have interacted with her without knowing. Maybe. So, so, so, um, yeah, we, we launched her and, uh, her goal was like to kind of create our own persona, being but pretty transparent about being AI, although still being kind of like— she talks about her being like the, the, the, uh, a blurred line between reality and, and virtual. That's how she describes herself. Pretty cool. Uh, launched like, you know, a few, few, uh, social media accounts. You already combined has 500 followers in, in 3 weeks. Um, all industry people, uh, creates really engaging content. Got me on a couple podcasts already without, without trying, right? Like, it's just got kind of like, um, uh, organically worked. And that was kind of our first experiment. And we're still running that. Honestly, it's like, I wouldn't even call it just an experiment. But what I think gave us a realization of like how crazy it is to give an AI agent full access to a computer, right? Browsing access. A memory access, email access. And the crazy thing is that the level of autonomy that it can do things, and me not being— I'm not a technical founder. So obviously we have my, you know, my co-founders, our CTO, Gio, is— he's very technical. But I was able to set up— I say you cannot be someone who doesn't know anything about AI to set this up, but you don't have to be a technical founder. And created API connections. Like I got her a Slack channel. Like got her to create like image generations like that with an identity that looks exactly like her and it's consistent across all our posts, which is pretty crazy, right?

Like not me. Are you happy for us to share? Because I know Sophia Vienna, you're right, on LinkedIn. So are you happy for me to share? My screen and show Sophia Vienna on LinkedIn? Yeah, please. All right, let's do this. All right. So let me click this. I need to click on this and I will click on, is it in Google Chrome? Yeah, it's in Google Chrome. Right. So this is Sophia Vienna. This is the AI agent or OpenClaw that Luca and the team created. And I, I know when I saw Sophia Vienna first, because I saw and I was like, oh, I didn't realize Journey had hired somebody new. And I was like, that looks— that image looks a little bit AI, but hey, let's run with it. Everyone's got these AI images at the moment. So talk us through how you created Sophia Vienna. And so you created a LinkedIn account. How did you create a LinkedIn account? Did you like set up her own Google account? Like, how have you managed this?

Yeah, so exactly. So basically we, we give her a company email Um, and then, uh, basically give it a company email, kind of creates— because it's under Gmail, it creates automatic verifications across multiple accounts. But she went— I, I set it up. The initial setup is a little bit like tricky. I think that's where you kind of need— I don't want to say like you need some coding experience because I don't have any, but I was literally like, uh, because I didn't want to bother my tech team And I was literally like screen sharing with ChatGPT on one end, showing what I was doing on the terminal on my, on the other end of my Raspberry Pi. And I kind of got it set up. Then I had created integrations with the messaging app Telegram.

Telegram.

Thank you. Um, and so I basically, my goal was I only want to talk to her through Telegram. I gave her Google Drive access so she can share files to, to, to Google Drive and so on. I said, go ahead and create your LinkedIn profile. She went and, and honestly started doing it.

So it set up its own LinkedIn account. So it went through, it went to linkedin.com, it created an account, did all of that. All of it.

Wow.

And then what else?

I never logged in on her account.

No. So only she can log in. What about Instagram, Facebook? Or has she just done X? Has she done all of that?

She, for now, we're only on Instagram and LinkedIn. Yeah. And she has a, she has a, Gosh, I forgot the name. What is it called where you share all the conversation?

Reddit. Here we go.

Reddit.

She has a Reddit account. She doesn't post on Reddit, but she, she goes and reads conversations. And actually, that's a funny thing, right? Like if you look at her content, which by the way, I don't control.

I know. But you control the content.

If you go look at her content, like if you look at the third post, if you go Click see all posts. I mean, if you go look, yeah, look at that post. Got 33 likes, 9 comments. Like, like she writes engaging content. That's, this is a, I mean, I don't know if people realize what it means that just on LinkedIn in basically less than 3 weeks getting 133 followers, but it's from zero, right? Like it's starting from zero.

It takes, takes a while. I'm just seeing if, um, I'm just seeing if there's any comments back from other AI agents. You know, the AI bots. I'm just looking. These do look like— these do look like a human. You know, you can spot a, um, a LinkedIn bot comment a million miles off. This is fun. Um, okay, so you chat to Sophia in Telegram, right? So when you're chatting to Sophia, like, what are you asking Sophia to, to do? Like, how are you engaging? Like, how is she engaging with you? Like, how are you like chatting backwards and forwards?

Well, I set her up to do activities 3 times a day on different social and send me a daily report on her activities. I told her certain parameters that she needs to escalate things to me. If somebody sends her a DM asking for sensitive information and so on, again, like, you want to treat this like somewhat careful because I think the cool and the bad of OpenClaw is that you can give it a lot of autonomy, but that also—

Yeah. So has Sophia had people messaging her? Because I can just imagine the sort of things that LinkedIn men would be sending Sophia in a DM.

But there has been people asking if she was AI and she's been very transparent about it.

Oh, good. So this is going to be my next question. Does she admit that she's AI or is she out there just catfishing some very lonely single men or women?

No, no, no, no, no. The goal is to be as transparent as possible. Actually, her Instagram account has it even on the bio. But she kind of acts mysterious. She talks about like how she's like the blur between reality and digital, like, and how the line is is getting thinner, very blurry, very blurry. Right. Yeah. But, but she, she, you know, we don't want to, like, mislead people. We said, like, be transparent about it.

So she's— Yeah, well, that's good. So those parameters are set. Now, here's the next question. Has Sophia generated any form of revenue or a potential sales call for Journey?

So she has. She got me on two podcasts and 3, and she got us 3 demos so far without outreach.

Wow, wow, wow, wow, wow, wow. And if they were to close, let's say these 3 demos, let's just say 100% close rate. Your sales team are fantastic. What sort of revenue could that put? And let's just say 20 properties in, in, in Manchester, right? Let's just say on average, that's the per client. What sort of revenue could that do for Journey over the course of a year if like the client, you know, stays for a year, does all the things Are we talking 4 figures, 5 figures?

Yeah, I mean, like, I think if she were to close even, let's say, I call it 50%, I know it's 3, so it wouldn't be 50%, but let's say 50% of the potential revenue that she created, um, would be close to like, uh, $25K yearly.

And then how much, so Raspberry Pi is one of the cheaper models that you could potentially do with OpenClaw. We, we had the person before saying that they could easily spend $600 a month if they were keeping on the Opus 4.6. So return of investment-wise, we're talking quite a lot here.

So, so there are like, we run, we run the models on like through API keys with, uh, we don't use Opus because Opus is insanely expensive. So there's, there's a lot of optimizations that you can do. Like, so it seems easy, but it's not easy. What I'm saying is like, there's a lot of optimizations that we know just because of talking to my tech team all the time. So like, and I was able to optimize her quite a bit. Like we started and she was costing $250 a day. That's not good. We dropped that cost to like less than $15 a day right now. Wow.

Through model optimization. And what else is Sophia doing apart from being business development? Or is Sophia now one of many agents or OpenClaw agents that you have got going on behind the scenes?

That's, that's what's happening. So right now we have kind of had this crazy idea. Should we give LinkedIn accounts to all our AI agents employees or not? So we kind of like, we're kind of debating this internally. Yeah. But as of right now, we have almost a dozen full agents within the organization. Wow. And Yeah. So considering that we're a pretty lean team, it's soon going to be probably 50% of our workforce will be digital.

That is insane. Well done. Well, well done. So, okay. I literally, I am stumped with this because I remember seeing Sophia a couple of weeks ago and thinking, oh, that's cool. Journey have got a new hire. And then Lo and behold, it was an AI agent. And it even is more crazy of it, uh, generating potential revenue as well. Good revenue. So now that you've sort of opened Pandora's box, where do you see this going now? Like, obviously Open Claw got the head start on agents, and obviously now, um, Peter Steinberg is being hired by OpenAI, so it's only a matter of time before other models bring out their own agents. How do you see this all playing out for the rest of 2026? What's like your Mystic Meg sort of predictions? You know, you've got the, you've got the predictions ahead of you. Like, what, what are you thinking happens between now and the end of the year?

Well, sometimes I just— we, when we stop thinking about it, it kind of feels like a bit surreal because there's so much happening so fast. Um, But I can tell you that obviously we are setting ourselves almost in every single industry for disruption, right? But it's also going to create some of the most amazing opportunities ever, I think, which is very interesting, especially, I say, for the short-term rental industry, but also like in general for— there's, there's a very key team size and stage of companies that are going to benefit from this the most. So when you're large and established is the worst place to be in this moment. That's— I don't want to be that. Yeah. But if you are anything between kind of like in a pivotal moment or even earlier, it's a great time to be. And what I think is also very interesting in the short-term rental industry specifically is that, um, these tools are going to enable small teams to do more at a much less cost than large teams. So this means that being small is an advantage. It means as long as you pay enough attention to what's happening around you. But it's an enhancer. It's an enhancer. So what this allows you really is to take a person and I don't see, although it looks like we are substituting people with AI, but that's not where we see it. We see it that basically every person has their own almost assistant that 10x their productivity. The fact that the people are saying AI is going to take jobs away, it's probably true somewhat, but I think more than that, it's going to increase productivity like crazy.

Like, yeah, I can, I can attest to that. You're right. I feel like if you've got a small team, like we personally have got a small team, every single— at the turn of the year, I upgraded every single person at our company to have a Pro plan, whether that is on Cloud, whether that's on ChatGPT, whatever. And their productivity levels is through the roof. Our customer satisfaction is through the roof. The products that we are shipping and the rate that we are shipping is crazy. And, um, that is very true. Empowering humans within your organization with the tools of AI, because we're still very early. I think there was a crazy report that 0. 0.04% of the, of the whole world have got a pro plan on AI, whether it's Claude, right? That is mad. So you are so far ahead. Even like every single one of our devs have got a pro plan or a max plan on Claude, and they are now ahead of 0.04% of the whole world, which is fascinating. So yeah, you've got to be small, you've got to be very, very, very proactive in this. You can't just dig your head in the sand and just go, I'll come back to this in 2027. And if you, if you can do that, then you're definitely ahead of the, ahead of the game, right? Let me get into the question that the guest before left you. So this was Clive Titmus from HelloHostie. He asked you, and he didn't know who you were, so I need for you to think of the question for the next guest who's coming in very soon. So the question was, what security parameters have you put in place when building agents for either your personal or for your business so they don't go go totally off the rails, you know, start doing all the crazy things, like what are you putting in place?

Well, it's a very good question. So in the least technical advice that I can give, because obviously there's some technical aspects that, that dev team can, can, you know, someone who's very tech savvy can, yeah, can take this much further. And we have taken precautions, especially on the product side, but I can say isolation, right? Like, why do you hear these people that buying an actual physical computer and put these agents inside of a computer? Because it's an isolated environment. You have a plug, you can literally unplug it, right? Like from the power, if for some reason something goes off the rails and goes mad. So as isolated environment as possible. Knowing, maybe sharing information, knowing that whatever you're going to be sharing, it could go out in public, especially if you're new to this technology, just thinking in these terms. And, but then you have two extremes, right? You can have more super isolated environments like a ChatGPT, you create your own GPT and you give it its own knowledge base. And you know, that's somewhat more safe. But then if you're doing something like Open Claw, you just need to know that whatever information you give it, there is a chance that it might get put out there. If you create an API key, give it a credit limit. If you wanted to give a credit card because you're crazy, make sure that there is a credit, that it's a digital credit card that has a daily limit or a monthly limit on that card.

That's a very, very good answer. Have you seen people that have been giving their agents to stocks and shares and crypto. Have you seen that?

I've seen that. And that's crazy. It's funny. Yeah. We've seen it. We're actually testing some of those things. I think it just, this is a complete experiment because we have this like almost internal lab that we experiment things with. I think it's a bit early on that because there's so many variables. It's like to do really, really well. But yeah, it's going to be interesting.

There's going to be tons of opportunities. All right. Very good answer. Have you got a question for the next guest? And you don't know who the next guest is. The next guest don't know who you are. Have you got a question to ask for our next guest?

Yeah. What have you had a ha moment that made you realize AI is going to change everything? And if so, what was that?

Oh, brilliant question. Have you had a haha moment and what is it? Right, I'm writing it down. Luca, an absolute pleasure. I'm looking forward to seeing you actually in person at some point this year, and I am definitely going to go off and chat with Sophia later on and see what I can get out of her, see what she thinks of Luca. All right, I'll let you go. To everybody else, please say thank you to Luca in the comments. My next guest is about to join any second now. So let's go have a little 30-second break. I'll come back with our next guest. Please go and check out Luca on LinkedIn and Sophia as well and go and chat and say hi. Thank you very much, buddy.

Thank you. This is a freestyle.

I'm going off the dome.

Ready for my close-up.

Good looking.

Let me tell you, get more bookings. Oh, that'd be nice, giving me tips, tools, and advice. Yeah, you're gonna get them, hon. You're gonna sit back, listen to Mark Simpson. Whoa, having a blast, gonna get it on the Bruce Lee podcast. Bruce Lee like Bruce Lee because it's so hard and the tea is loose leaf. Making up those rhymes, don't write it, just do it loosely. If you want my respect, you're better put direct. Here are the words in the podcast, that's what comes next.

I welcome back everybody. Uh, we've had two fantastic guests so far on this huge, uh, agent livestream, and I've got another fantastic it one with me. Conrad O'Connell, thank you so much for joining us. Um, so before I let you do a bit of an introduction, uh, what we've been doing today is we've been delving into OpenClaw mostly. We've been chatting a lot about OpenClaw, but what I want from your session is— well, we'll dabble a little bit into the claw, but I also want to talk about the other agents and the other platforms that people who are watching this with us now or later can, can dig into. So before we get going, uh, Connor, just give a little bit of background about yourself, about the business, and just sort of delve into your journey into OpenClaw, if you don't mind, to get going?

Yeah, yeah, for sure. So Conrad, founder of Buildup Bookings, we're a digital marketing agency. We work specifically with, you know, short-term rental vacation managers. Our focus is primarily on search, social, and email marketing. So that's where we spend most of our time helping clients get more, you know, traffic visibility. You know, direct bookings is kind of one thing we go down the rabbit hole a lot on. So that's— people can learn more, do some more digging on that if it interests them. I think to go to your second question, probably why people are actually listening is, I don't know, there's something about it. I don't know where exactly it is in my brain where I just look for the new thing. And for me, like usually that surfaces on what is now known as X, but what I still reflexively refer to as Twitter. So like I will just scroll Twitter and I'll just look for things and if I find it interesting, I'll dive into it. So I feel like that's where I've gotten, that's where I got the bug, you know, or whatever it was, or the, the original Claude bot name and now it's Open Claude or whatever. Where I first installed it back in January was off seeing a Twitter thread of like, this is novel, this is new, this is interesting. You should play with it. It has all these capabilities and I don't need much more enthusiasm, encouragement than that for me to be like, okay, let's try it. You know, like I just have that mindset a little bit with when it comes to software.

So I remember seeing your LinkedIn post a good 3 or 4 weeks ago now. And you put a post and a picture up of your, of your OpenClaw. So do you want to delve into the name and, and how you came to it and sort of like how you got started and how you you like cracking with it?

Yeah, in the terminal, like it asks during onboarding, like, what do you want to call this thing? Um, or maybe it's when you first set up the Telegram bot, I forget at this point, but I was like, I don't know what to call it. You know, I just thought of a silly name. And then of course, as soon as I posted it, people had better names, you know, in the comments underneath on LinkedIn. But, uh, I think I was, I was going golfing that day or the next day and I was like, all right, it'll just be a golfing lobster. Like no one else will have that.

Right.

So I called it Bogey Bot. So, uh, I like alliteration. If you, if you haven't got that yet, you can build up bookings.

Always nice.

Things.

Bogey Bot. So we got the Bogey Bot. So how did you set it up? Did you go and dust off an old computer that's in the office, or did you go and do the servers that a couple of people are doing?

How did you go ahead and set this up? Neither. And maybe I did some things that, again, you know, some people would say are riskier. So I'll admit that, like, you should do your own research and diligence. But I put it on my main machine, you know, and I didn't give it access to everything, but I put it on my main computer, which has mostly been a positive. I do think that There are some downsides to that that you wanna be aware of, particularly when it's doing browser actions. Like it can literally hijack your computer. So like you're in the middle of a meeting or like we're recording right now and if it were to go trigger a browser action, it would like actually open a tab and like, you know, the other screen over here and start doing things in the browser. Um, which is a little odd and futuristic when you think about it, but it does, it can like disrupt the actual workflow of what you're trying to do in the day if you put it on the main machine. But for me, honestly, it was just not having the desire to go and switch with, you know, mess with servers, configs, and all that kind of stuff. And I actually think having it locally on my machine is safer because I can actually sort of keep an eye on it and look at it every day, etc. And I just don't have another machine handy that I could put it on that I feel like would actually, you know, be ready to go. When I'm done with my computers, I typically sell them. I don't have a lot of backups. I have one backup, but it's in use by my kids. So I hesitate to—

So this is fascinating. So you give it to your main computer, right? So how do you decide on what it can and can't do, what it gets access and it doesn't get access to?

I was relatively— again, this is like maybe my personal tolerance for risk is higher than other people. I also think I have really good like operational security on things that I would, you know, be worried about it doing. For example, you know what I mean? Like I did, for example, I didn't give it email access. Um, so that's one line that I did draw where I'm like, okay, that probably seems like more about than good. Um, I have done that with other tools like Claude, for example. I have done that, uh, in read-only fashion, but I did not do that for Open Claw. But I mostly gave it access to things that are local on my machine. So obviously like file folder structures where we have reports or where we have, um, you know, different information. I did give it access to like Basecamp, which is our project management system. A lot of people use something like Asana or something like that, but that's, that's what that has access to there. And, um, transcripts. So like when we do calls, everything is transcribed. Um, the particular software we, we use for that was called Granola. I know there's a lot of others out there that do something similar, you know, transcribe. I think you've talked about some of them before. So that's, that's what I more so gave it access to because again, if you think about it like a personal assistant or something like that, you may or may not give a personal assistant access to your email inbox. That's more of a depends on what they're doing type thing, but they for sure would have access to like your to-do list, your calendar, you know, your meetings, um, notes that you have from those things, your task list. Those are things they would have access to. So those to me felt like the way to go. So that's where I, I started for sure.

And before we talk about Claude and the other agents that are out there now, how are you using it? So you've, you've given it access to transcripts, which is great. Uh, got access to Basecamp. It's got access to a few things. Like, how are you then taking all that information and how are you personally using it in, in your day-to-day?

Yeah, the first use case I had, and this is, this has been refined a little bit since when I first started with it, but I, again, on Twitter or X or whatever, I saw a thread originally where someone was like, okay, You've seen all the hype threads, like zoom out from that for a second, pause for a second, take a breath and say, if I was gonna hire someone tomorrow and they had, you know, I wasn't paying them anything basically, or I'm paying them virtually nothing and they could just help me with things, what would be the things they would help me with? And that reframing sounds simple. Like everyone would nod their head and go like, yeah, of course. Like that's what you should be doing. But then we get a little distracted maybe by like all the shiny object things that we might see online about what this thing does. So I said, okay, for me, what I often struggle with personally, and maybe I'm oversharing here, but I'll share it and let the consequences stay. Is just having the right context when I go into the calls that I have, because I'm often booked like back to back. Like it's not uncommon for me to be booked back to back on calls Monday, Tuesday, Thursday, like Friday, even some days. Um, so that's, that's a common thing. And I always feel like I never feel like I do enough prep, to be honest with you. Maybe it's just a skill that I've developed over 10 years where I don't need to do as much prep as I should. And I can get on the call and bring up numbers relatively quickly and dive through things and my brain can process them, but it's not optimal. Like we all, what would be more optimal is me doing more digging before the call and having some things you know, a little bit more prepared. And we have client success team members that help me with that as well. So it's not like I'm completely solo in this, you know, in the, in the nature of how my work works, like how my agency works. But yeah, that's the thing that I started with. I was like, okay, at least gimme like the last 30 days summary of everything that this client has been working on with us. Because to, to be frank too, I'm not privy to everything that goes on. Like comments may happen in Basecamp, projects may happen in Basecamp that I'm not looking at. So for me to get a little summary of that, that's just like, you're meeting with Mark tomorrow. In this case, pretend you're a client. Um, Mark is doing PPC ads with you. Um, here's the last 30 days of communication between him and someone on my client success team, or he sent 2 emails about this, he replied to it. And I just forget, to be honest with you. Like I just, by the time 25 days, I don't remember what we talked about 25 days ago, at least not to recall, you know, in that second. So I really just started it like a prep, like a, like prep me for the meeting, give me more, as much information as possible, summarize it, and then send it to me the morning of so I can read through that. Like it's called pre-reads, like I do all my pre-reads, which I actually stole that from like a pres— I guess the president has that. Something similar to that, or maybe in the UK, the Prime Minister or something, you literally just get a morning briefing. It's like, you're meeting with these 4 leaders today. Here's 9 things you need to know about each of these people and what's going on. And that's how they, you know, ask about their kids and know what they're struggling with and what's going on in their countries. It's not like they actually know everything. That yes, the people are, countries are very smart, but they don't know everything, right? It's that they get prepped, you know, beforehand. So that's kind of what it started for me.

Yeah, I was doing that. That's fascinating. And how are you chatting with Bogey Bot? Are you doing it on Telegram? Have you figured out something else? Like, how are you sort of chatting to him?

How did you do it? Yeah, kind of maybe this has been my story so far that you've noticed. I just went for the path of least resistance. So I know there are other ones you can set up. Actually, frankly, I had a Telegram account or maybe I'd registered one previously. I've never used Telegram really. Like there's no friends or family or team members that I message on there. So I put the app back on my phone and got it connected and just went for their, like when on their onboarding list, it was like, this one's the easiest. I'm like, cool, let me do that. Like I don't mind using a different messaging app. You know, I know some people use like Slack or other messaging things, but for me, I was like, let me just try the easy one and kind of see, again, look for the path of least resistance to get it going.

So that's what I chose was Telegram. Do you chat mostly on your phone to it? Or you got like an app on your computer open? Like, how do you mostly like engage and dial in with it?

It's probably 70% on my computer and then 30% on my phone. The phone thing is really helpful because I'm not in front of my computer 24/7. So like, it is helpful to be like, you know, again, I could go prompt it and ask it to do things for me. Again, think of it like a junior employee that will do things for you. Hey, I'm in the middle of meeting right now and Mark wants to know how his analytics is doing. If I was, if I'd given the right access, it could go in and say 30-day summary report of analytics. It could do that while we're chatting, that sort of thing. So that's when I use it more on the desktop. But yeah, the phone use is valuable because you can have it do things for you while you're away and then you come back and those things are in a somewhat complete form.

So if you're on the golf course and you're like halfway around and you're like, oh shoot, I wish it could do this. Boom, up you go, type a little message, voice note it in, and it's done when you get back.

That's nice.

Exactly. Yeah.

So, and clients who are listening, I'm never on the golf course during the week, of course, you know.

Of course, of course, of course, of course. And you've given it read access only, so it can't write anything, it can't change anything, but it gets to read a lot of things.

At least not my email. It can read into like, or it can write into Basecamp, it can write into like Calendar, for example, a few things there. But I, I did stop one step short of it where it can't write an email for me, but it can, you know, read the emails and go through everything in Calendar. It can go through everything in Basecamp, um, et cetera, and the transcripts. But it can't, the only thing it can't write into right now is email. But I don't know, I don't worry about that too much, to be honest with you, like at giving that permission because maybe it, maybe it's just me not being as concerned about it or you see a few horror stories, but maybe this is just my personal tolerance for like, you know, putting a little more risk into it. You gotta figure out what's fair to you or you want to do, but I wouldn't be opposed to it now, now that I understand how it works better.

Fascinating.

And this is what I've loved, is that we've had 3 different people on so far, we've got one more to come, and every single person has done it a different way. And that's fascinating. So there's more than one way to go about this. And the, the next thing I want to ask you about, and this is what I really wanted to home in with you, is about the different agents. We've obviously chatted about Open Claw a lot so far today, but there are ways and means that other companies are now catching up, Claude being one. So just break it down for us, the, um, Anthropic and OpenAI and Google. Just break down those companies for us. And then what have you noticed more with Claude ever since Peter Steinberg decided to go with OpenAI and not, you know, a Meta or an Anthropic?

Yeah, so playing with both these platforms now, the one I can feel like I can speak most confidently on is like the, the co-work platform specifically within Claude. Obviously there is like a Claude code it's probably a little bit more meant for developers. Yes, there are many non-developer use cases, but it's a little bit more intended for developers. I'm not a developer, don't pretend to be one. Um, so I, again, 90% of my use case of the Claude updates have been in the Cowork function specifically. Here's the analogy that I've been doing, or I came up with this the other day. I think it's the right one between these two platforms. If you're considering both of them, OpenClaw is kind of like a, um, a race car. Like it's, it's somewhat unstable. Like it breaks sometimes. It doesn't always work very well. Uh, it's very expensive, by the way. We haven't touched on that yet, but if you're using OpenClaw a ton and you're using the latest models from OpenAI or from, um, Anthropic, you know, using those, those high-value tokens, I mean, I easily burned through like 40, 50 million tokens in the first few weeks getting everything set up and, you know, getting the hard—

What's that in dollars? You say 40, 50 million tokens. What's— we sort of put in like a dollar amount to that?

Over $400 of tokens from Anthropic, like easily over $400 of tokens. Yeah. So like it was burned a lot, you know, and, and I knew as it was happening and then I'm looking at it and like, again, so to do my race car analogy, the race car burns very premium fuel. That's very expensive. You know, it kind of breaks a lot. You gotta have a crew of people there. That's it. That kind of thing. That's kind of how I feel with it. And then cloud to me, coworking in particular feels like a Mercedes-Benz, you know, or BMW. It's a very premium thing that just like everything works out of the box. Maybe it's not quite as powerful as a race car fast. It's not that., you know, raw on the edges. It doesn't do the same sort of unhinged things that you could try with OpenClaw, but it's a much more refined sort of experience. Like, it feels very predictable when you're asking to do things. Um, and that's honestly where I've— some of my use cases have gone to is like, well, I want to polish presentation. OpenClaw— or excuse me, ClaudeBot— excuse me, let me try that again. Claude Cowork is the one. There's too many names here. Yeah, more than doing that, a lot of my like day-to-day like quote unquote knowledge work has really transitioned into that platform. And they've added some things over the past, like literally the past week as we're recording this, um, that are really moving more in the direction of Open Claw. So for example, Cowork added scheduled tasks a week ago. So those pre-reads, I'm now realizing that I can do them. And I set up one to run in both systems actually this week to kind of compare and contrast between the two. And honestly, I'm really liking the ones out of Cowork better than the ones out of, um, Open Claw. So like that already, that one I'm turning off Open Claw and focusing on Cowork using that pre-read system.

So with Claude, you can connect that up to Basecamp, you can connect it up to all of the places that you already were connected in OpenClaw.

Unfortunately not Basecamp. They don't have a pre-built connector for Basecamp, maybe just because it's a less popular, you know, project management software. They do have other ones in there like Asana and I think Trello and some other ones that are quite popular, but they don't have Basecamp in there. Um, I did, I did end up finding like a third-party sort of connection, you know, with Basecamp, but it wasn't an official one like in the Anthropic, you know, yeah, because you can do it via like, um, Supabase, or you can do it via like a third party.

So you can go, yeah, um, because I've just done Airtable, Airtable into, uh, Claude or wherever via Supabase. So it pulls it all and it moves it around and you still get the data, it just got to go a little bit of ways to get it.

Yeah, yeah. So, but whereas Open Call, like, it— this is the difference again between the two. Open Call was like it didn't exist. There was no pre-built, obviously, connector for Basecamp at OpenClaw, and there it was just like, oh, I'll build one. That was one, that was one of the things that spent and burned a few million tokens. It was like, I was like, cool, sounds good, like, go build it.

Yeah, you can build it, but it's gonna cost you. That's fascinating. Yeah, yeah, yeah. But that's good. So when you see people talking about, uh, on X or Reddit, and you're talking about OpenClaw, and they've got these things called a marketplace, and you can go and download skills per se, is that people that have created their own skill? Like, you connected up Basecamp and then they've gone and put it public. Is that what we're talking about when people figure that out or is that something else?

I would differentiate between like a connector and a skill. Those are two different things, right? So like a skill is really more so just similar to like how an OpenAI project works where it's like more so a set of instructions for the Claude to follow. That's how I think about the skills. So it's like when I'm like the most recent, the podcast that dropped literally two you know, as we're recording this basically was Paul and I did an episode on how to build a homeowner landing page. And we came up with these like 20 different ideas of like sections for a homeowner landing page, right? So that I've now actually turned into a skill in Claude. So I can then prompt Claude and say, use this skill that I developed and build a homeowner landing page for whatever, insert vacation company here. And you have to give it the right context, right? You gotta paste in the, some information about the company, the name, et cetera, et cetera. But the skill is just a preset list of instructions to say, like, just like you would give an SOP to a team member, that's kind of what a skill is.

More smart. Oh, I like that. Skill is an SOP. So technically now, if you wanted to, you could just go and get your Claude or Open Claude to, let's say, go to Myrtle Beach, go and find 10 property management companies in Myrtle Beach, and go and look at their site, pull the data, create a landing page, and then could either LinkedIn message them or email them, say, hey, I built this for you for free, let's chat if you want to chat more. So you could do outreach with it, really.

Yeah, I think, I think I haven't gone to the last few steps that you've referenced, but the first few steps that you referenced sound very feasible to me that would work right now out of the box. The last few steps may require a touch more, again, sort of like engineering or kind of like breaking it into smaller bits. Like, okay, let me do the outreach portion first. Let me then do the page generation second. Let me then, or excuse me, let me do the, like the finding, the scraping is kind of step one. The collection of the pages is sort of step two. The creation of said pages is step 3, then the outreach is step 4. So it could be something where it can't just, you know, in, in the Twitter world, you'll see people say like it one-shotted it, which is like you just put one prompt in there and it just does the whole thing. That seems like it could pull that off to me based on how you've described that. But I think if you broke it into 4 sections or 3 discrete sections of what you're doing, I think that's absolutely feasible. And that's where, again, I think these systems still today very much need what I believe, like a lot of people would talk about is like human in the loop. I don't know if you've seen that terminology before.

Yeah.

But it's like, okay, at some point it's gonna stop, it's gonna finish. You're then there to do that next step, you know, and say, okay, how do we get there? And maybe this is just my own personal limitations, knowing what I know and knowing what I don't know. I think that's kind of a lot of where I find myself is like, okay, there's points where I have to stop, give it additional feedback, give it additional context, tell it to do the next thing or move it from one skill to the next or one thing to the next and say, okay, now that I've got this, let's start the next thing. I'm not seeing as much as people claim that like you can just click one button or do one thing and it's this autonomous thing that just goes end to end. I'm not seeing that.

Seen that as much. It would be nice to have that next— that's the next level where you can just be the— so BogeyBot is the overriding person, and then it's got all these different things happening between it, and it can just be run on automation. Because I'm still struggling personally to figure out where it becomes a— oh, that's a nice to have, as in like it's pulling in my meeting notes, it's giving me like a little debrief before I chat to a client or a prospect, etc. Where it's genuinely generating money, um, and where it can start to do things like outreach like that. Where if I was to pay a human to do that sort of outreach, we're talking a lot of hours and a lot of money, right? Probably $1,000 to $2,000 for an appointment setter to go and do that sort of thing without even scraping together a website. But if you can get someone to sort of— or something to do that, where it's getting in front of people, but then they go, yes, I want to book in a call, then that's pretty decent. The guest before your good self, um, he had created through his PMS— so I won't tell you who, but with his PMS he has created, um, uh, OpenClaw, and he has created, uh, a LinkedIn profile, and it was a LinkedIn business dev. And we've just seen a live screen share where we— this OpenClaw called Sophia has gone in and has actually booked in free demos for their sales team, which I find is fascinating. Like a LinkedIn post has got, I think he said 20 likes and about 30 comments. So that's the part of all of this where I'm thinking, right, as a business owner, if I'm going to sink time and energy into this thing, I need for it to actually have a return of investment and not just me messing around with a few things. And it's like a, it's a nice to have because I'm taking away from the core reasons to run a business, you know what I mean? So that's where my head's at with all of this.

I'm not going to challenge you per se on that, but this is just my counterpoint to what you're saying there a little bit. I think that you don't know what it can and can't do until you get your hands on it, right? It's sort of like, I think the best thing for people to think about with like the new frontier AI technology is like finding a use case. And I think you actually have a really good use case, to be clear. Like your thinking is spot on in terms of like, this is where I need to go with it. This is my ultimate outcome from it. But then it's like the way to get there is gonna be a little messy, I'll be honest. Right? It's not gonna be like, I'm gonna set this up in one, one-hour session and then I'm done with it. Right? It's gonna be like, just like a person, right? You have to, you have to train a person over hours, tens of hours, hundreds of hours. Right? There's people on my team that I've spent 100 hours with on phone calls and I feel like I still train them something new every week or every month where I'm like, oh, you didn't know about this. I taught you about this. That sort of thing. So that's kind of my like mental understanding of how I think about using these AI agents is like, It's not gonna be one-stop shop. It's gonna, but you have an end goal and that's what everybody should have, right? When they start this project, I know what I want out of it, or I know the problem that I'm missing. You know, I feel like I'm missing right now. This, these are the steps to get there. And then you're building, you know, each thing block by block, you know, Lego brick by Lego brick until you get where you wanna go and expect it to like fail, like expected that there's gonna be some, you know, wasted effort and time. And then you screw up and have to delete the whole thing and start over again. But then when you're done, like the, the promised land on the other side of this whole thing is exactly what you described. Like when the agent is doing valuable work for you and it's costing you $50 a month in tokens and it would cost you $2,000 to do that with a person and it never is sick, never calls out, never has any problems, etc., then you're reaching that level.

Never comes in hungover. Yeah, all of those.

Yeah.

Um, okay, so you said earlier that if you see something cool on X, you want to give it a go. Have you dabbled with anything else? Because obviously like Meta have got Manus, um, obviously, uh, OpenAI from— I've got a ChatGPT. Have you been dabbling with any any other of these tools? Because personally, I've been dabbling with Manus quite a lot. Um, have you, have you delved into that world outside of Claude or OpenClaw?

The only thing that's— the answer would be no on the Manus side in particular. Um, the only thing that has me interested in that is the fact that it can do Facebook ads interface interfacing, obviously, because now Manus is owned by Meta, by Facebook. So that is one use case that I think I could personally find myself into. But I guess again, I'm As much as I like to hop on the new thing, I also try not to overly get distracted, right? I try to timebox this kind of stuff and say, okay, let's not do 9 things and have 9 things be half complete. Let's try to get good at 1 or 2 things, um, which is how my mindset goes. But the thing that would appeal to me about something like Amantis would be that, you know, with the fact that it has Meta access. But again, Meta would want that for other platforms too, right? Like, and presumably they will build some kind of connector so that people can spend more money on advertising, right? Through their Open Cloud install, through their Cloud, you know, Cowork install, et cetera, et cetera. So. So my belief is that will come. It's just they built it natively for their own platform first.

Yes, I'm sure I saw somewhere that they've built one for Anthropic, or they're literally gonna have the Anthropic one coming out very soon. And, um, I was one of the first Manus users. I jumped on that very shiny object and I used it quite a lot. Where Manus reached out to me, and I've just got on a call with them this past week, and I was chatting with their team in Singapore about the things that I like and don't like, and they said that the, um, the WhatsApp WhatsApp integration. So their version of Telegram is coming very, very soon. So I feel like they are getting more into the meta world, Manus, and fascinating to see what they're doing.

But yeah, there's going to be competing agent systems just like there's competing operating systems, right? There's Android, there's iOS, there's Windows, there's macOS, right? Like the same thing I think will occur here. And I think that it's silly to think that there's going to be one winner. I think there's going to be likely multiple winners, or depending on the use case, maybe prefer one or the other. Like, for example, you mentioned about different AI platforms. I don't find myself going to Gemini a ton now. But when I use Gemini, the image generation tools inside of Gemini are, in my opinion, the best. Like, I don't think you can make images that look on brand or match the right color scheme or fonts or whatever you're trying to do better than with like the Gemini AI Studio photo generation tool, like the Nano Banana stuff. Yeah. And so any, any other image tool I've tried other than that has been pretty okay to best, if not poor.

It's been so quick moving because at the start of this year, everybody was talking about Google. Everybody was talking about Gemini because they upgraded it and obviously like the Nano Banana and stuff. And now it seems like everybody's just gone so far away from using Gemini and they've gone like, you know, they've gone bouncing from, from, from one to the other. And you're right, there's not going to be one winner and there's going to be lots of winners, which is only going to be good for consumers because it means the cost will hopefully stay relatively low on that as well. Right.

So the guess— I think they've already gone down.

Yeah.

I mean, Google had some stuff yesterday where they did like a 3.1 flash or something like that. That's like It's like 80% as smart, but you know, 10% of the cost or whatever.

Right.

Like it's not quite as good as their newest stuff, but it's both faster and cheaper. So I think you're just going to see that, that token cost just continue to go down, down and down unless a model actually has something unbelievably unique about it. Like, oh, this is the only model you can use to make images that look like a person. So therefore they can charge a premium for that. But it's tricky to know too, because they're still losing money on all this stuff. Like I'm paying Anthropic a lot of money in tokens and subscription fees. And despite that, I know they're losing money on me. So that's the tricky part.

We don't know where it's all going to land. Yeah. Yeah, and there's obviously— there's a big huge Goliath that's just sitting back and watching everything, and that's Apple. So we don't know what Apple's play in all of this is going to be just yet. So your question, uh, the guest before you has left a question, and obviously after you've answered this, I'm going to get a question for you for the next guest. The question was, what was your aha moment when getting delved into these agents, and whether it's OpenClaw or Claude, and can you describe what that aha moment was? Do you have like that sort of, wow, this is pretty cool?

Yeah, I think for me it's, it's, um, well, actually I could think of one yesterday. It's maybe good timing on that, which is I was doing the, the pre-read and it went into my inbox and found a meeting had been canceled and already suggested what the next steps were based on the fact that the meeting had been canceled. So it's sort of like the proactive nature of here's what happened. It knew about it before I did because it went and read my email before I read my email. And it was actually giving me the right suggestion on here's what to do, here's how to handle this situation, you know, and go from there. So that, that to me was definitely an aha moment on the proactive nature of where these AI agents can go when you give it the right context, when you give it the right information about what you're trying to achieve, you know, your business, all that kind of stuff. And to me it's like, it, it felt like a moment where, wow, the AI agent is smarter than me, which of course they're all smarter than me, but it's actually more, it's smarter and more personal to my situation than I thought of. Like it took me. It would have taken me some time to come up with the steps of this meeting was canceled. Here's the problems, you still need to do this, you still need to do this. It sort of started on that for me and then gave it back to me and said, here's what you need to do now with this time that was canceled, meeting was canceled. Now you need to deal with this problem this way. So that was very what, what just happened kind of moment for me.

I mean, there's a lot of people who are on calls as much like you said that you are that will resonate a lot with that, which is good. Before we get your question for the next guest, where do you see all this going over the course of the rest of 2023? 26? Like, who and how and what are your sort of predictions with AI? And particularly, like, when we delve back into our industry of short-term rentals, like, what's your Mystic Meg sort of crystal ball view?

I, I feel like it's hard to do because if you asked me that a year ago, I wouldn't have predicted exactly what has happened, right, in terms of these, you know, AI agents. I, I think it's reasonable to assume that maybe the pace will slow down a touch, right? I feel like right now there's sort of this mindshare battle of, hey, we want to add in 1,000 functions At some point that mindshare battle will be more about like where the winners are, like what platform ends up being the most polished, like what do people actually end up using day to day and stick with for a little while. I don't think this idea, like you said a second ago, of like swapping from model to model and system to system every 7 to 10 days is actually a good thing, to be honest with you. It's probably a little bit more of a distraction than anything. So I think hopefully we'd have a good idea or a good understanding on this is the model that, you know, like there's 2 or 3 that are kind of winning the war, you know, on attention. You're using those regularly. And again, I think at the end of the day, we said this before we hit record, it's just software. Like it's just software that you're interacting with in the same way that a PMS is a software, an operating system is a software, Chrome is a software, a project management system is software. So I think that it will just be to the, like, the transition we talked about before we hopped on was this idea of when I started in the industry in like 2014, there's people that were ditching on-premises PMS systems. So it used to be like, oh, we had a server in a closet somewhere. Running a PMS software, which if anyone hearing that today, they would just laugh at that. That's like absurd, right? But that was happening back from like 2014 to 2016, '17. There's people that were literally getting rid of on-premise software and going to the cloud. And now like no PMS that was launched today would describe themselves as a cloud software. They'd be like, well, of course it's a cloud software. It's like assumed, like you just, of course that's how it goes. I think that same transition will kind of occur to some degree where it's like, of course there's an AI button here. Of course there's an AI interface here. Um, we won't even call it AI. We'll call it like like you can see this in Gemini a little bit, like a summary. This is just a summary of this email thread. They don't even label it anymore, but like this is an AI summary. That's a summary, you know, of it. So I think you're going to see a little bit more stability probably by next year. I think you're going to see a lot of different— I think the people that are most harmed by all these tools are the low kind of skill, low wage workers. So I think they're going to see the worst punishment, um, particularly, you know, if you're a VA and you're overseas, let's say in Asia or something like that, I think you're in a tough spot because your work can now be probably done by one of these agents at less cost and less headache, and it's more controllable, and there's lots of pluses there. So I worry about those people. I think anyone in the next year that's repelled by these things, that's like almost rejecting them out of spite or out of premise, oh, I'm not going to use any of the AI tools, I don't want to do them. I feel like they'll be very much left behind by people who are using them. And at some point they will succumb, you know, and they'll be like, I have to use these tools or I'm going to be, you know, very ineffective at my job. And doing what I need to do. Um, but that may take, that may take more than a year, to be honest with you. But I think by the next year it'll be very obvious, like, team members or people on your team that are not embracing it, they're, they're not going to do well. Like, they're going to really struggle. So those are some thoughts I have.

I have more, I guess. There's a, there was a crazy stat, and I put it in the WhatsApp group chat for the geeking out on AI, and I feel like it, it was something along the lines of 0.04% of people have got a paid Claude account in the world, right? So if you've got a paid cloud account, even if you're just paying them $20 a month, you are already ahead of so many people. And it's just time to take advantage of it. You cannot be burying your head in the sand anymore. Like 2023, when it first came out, people were burying their head. 2024, people burying their head. Even last year, people were burying their head and not doing it. I think now is the year to really do something because you can just— there's so much advantages to Do you see a scenario where somebody builds their own PMS with AI and they can literally be a one-man band? Like, is that something that you could see they could do?

I doubt it for a few reasons, which is that the variety of connections on the PMS layer is so vast, you know, so like, if someone had built quote unquote vibe coded a CMS or built a CMS with one of these AI tooling, I think that it's all good until you start to chat with the OTA platforms and say, okay, I'm one, I have one person on this PMS and I want to connect to Airbnb, right? Or I want to connect to the, you know, Booking.com or Vrbo or something like that. And they go, well, maybe like, maybe not. Like, why would that benefit them versus using these prebuilt connections or APIs that they've worked with for a decade with someone like, you know, insert 1,000 PMS names here that are out there. Now, will you see some level of, um, this is the thing too. Like people, people were talking about this and I talked about this a little bit with like WordPress kind of leaning into doing more on And, you know, people are like, well, why would I use WordPress when I can build my own CMS in an afternoon and it does what I want it to do and doesn't have extra WordPress kind of junk or stuff that I don't need. And I'm like, fair, but now WordPress is going and they're built like they have a huge head start. They have a 10, 20 year head start on building a CMS. And now they're, they're open source. We just have all access to all these same things and they're building on top of them. So that's what I would say is like, let's say you're going to go try to rebuild Guesty or rebuild Ownarest or Hostfully or whatever, pick, pick whatever PMS you like there. Well, those people in those companies have the same tools that you have. So they're already 10 years ahead of you and they're gonna keep adding more features and more functionality and go from there. So I, I really doubt it because I think that if you could actually do that, if you could build a functional, valuable PMS, you know, with a one, in a one-person environment, um, the best thing for you to do would be to sell it, not to use it for your own personal use case. Like all the people that the, the, the 99.46% that you mentioned a second ago that might need that PMS that are not using these tools are gonna be like, Well, how is it better than insert company name here, that sort of thing. So I'm kind of skeptical of that. But I do see people building tools on top of it. So like, you know, insert PMS name here doesn't have a cleaning module that's very good. I'm going to take my insert PMS name here and then build a cleaning module on top of it that's way better, but not rebuild the whole stack, but rebuild like little bits and pieces of the stack that they need in their business.

Add on extra things that will just talk to it and not replacing it, just adding on to it.

It's gonna be because people always have complaints about their software. Yeah. They always like, or let's say there is a housekeeping module, but they don't like it. Oh, you guys don't have a function for XYZ thing. Or, you know, with, with the, the template websites, as you know, you know, you and I have rallied against those in the past, you know, recently where it's like, okay, well I actually was shown a demo by someone who sort of quote unquote vibe coded a, you know, his own website on top of, he uses Guesty, like the Guesty API. And he said, yeah, I did this in like, 2 or 3 days and it's pretty functional. Like it's already there, you know. So, but again, that's not him replacing Guesty. That's him replacing his Guesty template website and him not wanting to pay a fee to hire Boostly or to hire Buildup or whatever the case may be.

So he can stop that. All right. Let's just calm down him, that person, whoever that was. Let's calm down. Don't tell anybody about that. That's like, it's not business model. Yeah. Fascinating. I've already said the website industry is screwed sooner or later. So fascinating times. Conrad, As always, I love chatting to you because you can take something really complex and just explain it in such a simple light so someone like me can understand. Thank you for coming on today and, uh, chatting about this. Really, really appreciate it. Now, before I let you go, do you have the question for my next guest who's going to be dropping in in any second now?

Yeah, yeah, it's, uh, I, I wasn't expecting this, so I feel a little flat-footed on it too, uh, to some degree. I guess my question for that person would be What are all the— what do you think all the benefits will be, you know, let's say from, you know, this new AI agent revolution? And we can do the same thing a year from now, that's fine. I can steal that bit. But it seems like a lot of the predictions out there, people want to focus on all the negatives. I would love to hear that person, the next person's perspective on all the positives. What are all the good things that are going to happen from this? Um, how are people going to benefit? How are people going to get—

and that will be a nice way to end this live stream, the benefits, because yes, we can easily fixate, and all you have to do is go to any Reddit forum or X feed and see the fear that comes with all of this— job losses, etc. But what are the benefits? And that'll be a lovely way to end it. Conrad, thank you very much. Everybody, please thank him in the comment section. Uh, go check out Conrad O'Connell on LinkedIn. Uh, some really good funny posts and just, uh, uh, just all-round great follow on LinkedIn.

Thank you very much. Appreciate it.

Have a good one. Hello, I am Chris Turner and this is a freestyle rap about why if you are in the hospitality industry and your website is not generating bookings, you need to go to Boostly.co.uk. Let's drop a beat, let's tell you why. I make it up, no script being said. I'm saying gonna get more heads on beds. Yeah, turn up the gas, getting it cooking because you're gonna turn the lookers into the bookings. You don't need Airbnb. Airbnb, they will go direct because currently they just step away. They're not staying on your website and so you don't get the top cash, but that's alright. You can go to Boostly.co.uk, money-back guarantee, and the clickers will stay because they're the market leader in web design for the hospitality industry. My mind to yours, and let me say, I hope you're enthralled. I make it all up. Go and book a little or call, and you'll be happy that you did. You'll be like, this is amazing.

Thank you, Chris.

Talking about myself in the third person there.

Yeah. Okay, everybody, welcome back. I've got my final guest of our very special AI Agent live stream. I don't know if you've done many podcasts before, but I'm really fascinated to digging into having a conversation with you. Mr. Hanno Dotzol, thank you so much for doing this. Before we get into all of this, can you just please let us know where you are in the world, how long you've been in the property game, and then just go into very quickly where you are with AI agents and how you got started, and then we'll get cracking with our little session.

Awesome.

Thank you, Mark. Thank you for having me. It's certainly an honor. I'm in Wayne, Pennsylvania. My properties are in beautiful Cape May, New Jersey, and Cape Coral, Florida. Hence the company's name, Cape Del Mar LLC. So I grew up in the '90s, big fan of Café Del Mar. For some of you, you might remember this great music which came out of it. And I decided to name it Capitelmar LLC. So how did I get into the game? I basically stumbled into the short-term rental game by just owning a second home, renting it occasionally out on Airbnb, and basically I was overwhelmed with the success we found. So we basically decided that we actually cannot afford to go on vacation in our own rental because the opportunity cost was way too high. And from there we expanded. So we are now looking at 6 properties, roughly half a million in revenue. And yeah, to your question, to directly go into this, how did I get into Open Claw? I was actually on a plane back from the Bahamas. And I listened to a podcast where they were talking about this new Open Claw setup. And I believe it was like 3.5, maybe 4 weeks ago. So relatively early on. And like, I was so fascinated with this entire idea of having your own agents, which are completely unhinged. Because I've been using AI quite a bit, but there was always this limitation in terms of what an agent could do, how long it could perform a certain task. And everything I've heard on this podcast really fascinated me. So from the plane, I ordered a Mac Mini, which got delivered to my door, I think, the same day from the KOP Mall. And I didn't get much sleep that night because I was directly hooked. And Mark asked me earlier whether I'm a technical person. I am not. So I have had to really figure this out all on my own.

There's a really nice recurring theme out of all of the guests, and I haven't mentioned to you who they are, but every single one of them has described themselves as a non-techie founder. Which is good because there's going to be a lot of people watching this with us now live, or maybe later on the replay, who are thinking, oh, it's all well and good for these chaps that have been on the podcast because they've got a technical background, you know, they know how to code, and none of our guests have today. So it's fascinating and really good to hear you, you say that. So you got the Mac Mini, it was delivered while you're on a plane, which is amazing. You didn't get much sleep, How did you get started? Like, what was the first sort of things that you did? And the question I've asked to all of our guests is, what have you named your OpenClaw agent?

Yeah, that was the fun part, right? So I created— I followed a setup which I think most of the people have implemented, where I have a chief of staff.

He's called Luke. Nice.

I like it. I think I was somewhat biased from, like, maybe Luke Skywalker or someone. But someone who could basically oversee the other agents and for quality control and most importantly, safety. So if you watch those, you know, YouTube videos, there are people talking about prompt injection. And of course, that is something which is pretty scary. So I basically asked AI all the time. Nothing which I really did was without the help of AI. So when I set this entire thing up, I had Claude, the add-in in my Chrome browser, running all the time. And I said, look, for instance, I give you an example. I want to ensure that I set up a system which is not prone to prompt injection? What are the terminal commands? What is like when you go through the OpenCloth setup? What is the important thing I have to pay attention to? And then basically Claude guided me through that. I mean, I still was somehow in the driver's seat in taking the decision, but I had very strong guidance in doing so. So I have Luke, Chief of Staff, and then I have 5 agents as of now. It's growing. I have like Tim, who is my personal executive assistant. I have Jeff, who is the STR, the short-term rental operator. He's basically doing everything around short-term rentals. I have Elon, who's the business development guy. I have Claire, who's my legal counsel. And I have JP, who is my financial analyst.

Luke, Jeff, Tom, Claire, Elon, and JP.

So we've got a team. Tim, not Tom.

Tim. Yeah, Tim. So we've built a team here. And what does— what— and so Luke is like in overall charge, is that sort of right? So he's in charge of Jeff, Tim, Claire, Elon, and JP?

How does the dynamic work? Look, they sometimes, they talk to each other. I mean, there are some really fun use cases which I've been working on. I think I'm going to give you one which I was somewhat blown away with. I, at one point in time, I started very slowly, mostly for because I'm a little bit risk-averse, right? I didn't want to hand those agents the nuclear codes right away. But what I at one point in time did, I gave Tim as my executive assistant his own email address, like a Gmail address, you can name it whatever you want. And I also gave reading permissions to my emails. So I got an email from a realtor basically saying, hey, I have this off-market short-term rental opportunity. There's someone who wants to sell his cabin, you know, in the woods in the Poconos in Pennsylvania. And so all of my agents are connected either through WhatsApp or Telegram. And I asked in this particular case, Tim, to analyze this short-term rental with all the information the realtor has sent in and also autonomously follow up with the realtor because she had only sent the P&L from 2023. So what Tim then did is he contacted the realtor, introduced himself as my assistant. He didn't reveal that he's an AI agent.

Yeah.

And lo and behold, the realtor answered that email saying, oh, nice to meet you, Tim. I'm going to get you the latest P&L shortly. So she didn't have that on her fingertips, at her fingertips, but she sent that a couple of hours later. Tim took this new information, processed it, like he went on to AirDNA, got comps, and then did a full pro forma and sent it to me and my investment partner. And he did this all like with like one prompt. I basically said, look, and it wasn't the perfect example. You know, I do not do business with this particular realtor all the time, right? So I thought she's good for an experiment. I hope she doesn't listen to this. And he turned this around full circle. I think this is a very good example how powerful it could be. And I say could be because there are only two limitations when it comes to open cloud. One is your imagination and how risk-averse you are. Everything else, you have to really think about like a person sitting in front, and we most of the time do this, right, in front of a laptop. Everything we can do in front of a laptop an agent can do, and you can delegate that to an agent.

So how are you now communicating with all of your agents? Are you doing it mostly from your phone with Telegram, etc.? Or are you— do you chat a lot on computer? Like, so how are you sort of working on your day-to-day with these? That's a phenomenal use case, by the way. But how are you now mostly communicating with your agents?

I think the great thing is, as you said, you have those both options, right? You have, if I'm really doing somewhat of a, you know, new project where I'm going to set up a new social media cycle and instructing the agent to do all of that on his own, in this particular case, I typically work directly on this Mac Mini. But, you know, I'm traveling, you know, I'm at my son's soccer game. It's called soccer here, I apologize. Yeah, we let him get away with it. Right, right, right. And this is the great thing about, you know, if there is any kind of idea popping up and you are indeed not in front of your computer, but you're out, you can just type that into either Telegram or WhatsApp. And it's all interconnected. There is this memory file, so they will remember all the conversations you've ever had with them.

And you mentioned chatting on the Mac Mini. So most people have their main computer, and then the Mac Mini's to the side in part of their office, etc., and they don't really use it. Are you going in and using the Mac Mini when you are chatting to it? So you're going away from your main computer setup to the Mac Mini, or how does that sort of work when you're, say, you're chatting on a desktop, you're in your office? How is like that sort of physical setup look like?

Oh, the physical setup is my Mac Mini is to my left and I'm looking at my laptop and they are somewhat connected connected because it's all this Apple infrastructure. So I wanted to have it somehow to a certain extent isolated, hence me setting this all up on a separate computer. But you can certainly also run this on your existing one or on a virtual server.

So do all of your agents have their own email account? With a Google account, etc., or is it just TIM?

Not yet. So as of today, only TIM has its own email address. That's just me not having had the time to set it all up because there's still certain things which you have to go through, right? I mean, the authentication, that is all stuff which you have to do once, and then the compounding, the recursive learning is hitting. But yeah, there is, I want to say, a good amount of work needed to initially set it up.

Yeah, that's been a bit of a theme as well for our guests. So you said you've got 5 use cases. That's one.

What's another use case that you have gone, wow, This has been, this has been great. One which I really got enthusiastic about for two reasons. One was emergent behavior, which I actually posted on X about. So the task was I always wanted to have all the conversations I've ever had with all of my guests somehow in a consolidated form so that I can properly analyze what are the top issues, create an FAQ out of those conversations, whatnot. And for whatever reason, the PMS I'm using is— that's not yet a feature. So I basically asked the agent to go in and scrape those 400+ conversations out of Hospitable in my case. So there is like a browser relay. So the agent goes into Hospitable and starts to scrape every single word you have ever exchanged with your guests.

Okay.

So after a while, I got a little bit suspicious because I was checking in and seeing on my browser that nothing happens. And I asked the agent what is going on. The agent was saying, oh yeah, it really takes a very long time, like 2 minutes per conversation. If you extrapolate that to 400, it's gonna take me 4 to 6 hours. Are you really sure that you want to do this? So he started to negotiate with me and proposed like, oh, I can summarize this conversation. I said, no, I have to have the raw data. To basically do a proper analysis. So he basically showed like emergent behavior, like a kid skipping homework.

Unbelievable.

He pretended to work while he was actually not working. So I got my chief of staff, Luke, involved and asked Luke, to enforce compliance. So then Luke started to talk to, in this case, Jeff, saying, look, this is unacceptable. You have to follow those commands. You have to follow Hanno's, you know, whatever he tasked you to do. And he apologized. It's not like that he was saying there was a technical issue or something. But he admitted that he was trying to cut the short and try to be more efficient and whatnot. So a little bit scary, very interesting emergent behavior. But eventually, my chief of staff got him compliant enough to fulfill this task and get the work done. And this was a treasure trove. I mean, like, he automatically identified the top 10 issues. He also analyzed that for certain issues, there was a kind of compensation on my end, which still led to a 5-star review, right? So let's say there was like, I don't know, something leaking, I don't know what it is, right? But he basically analyzed that through my conversation with a guest and certain make rights, I was still able to get to a 5-star review. And he very smartly analyzed those 400 conversations and gave me really good guidance on how to communicate with guests, what kind of amenities I have to maybe focus on and what to fix, you know, smart lock integration and whatnot. So the intelligence I got out of that was just mind-blowing for my business.

It's crazy to me how the agent tried to get out of doing the homework and you had to get your chief and staff involved to say, hey, come on. And then the agent apologized and then obviously did the job.

It's crazy.

So yeah, look at my X post. I posted some screenshots. The conversation between those two agents is just hilarious. And you really think— this is how you almost have to look at it. It's like they are born with some personality, but then you train them and they develop a certain personality.

Um, for those who were wanting to go find the, the post on Twitter or X, what is the— what, what is your handle? How could somebody go and find it? I'll make sure, but we put it in the comments later.

Yeah, I don't even know that top of my head.

I'm going to send it to you. You send it to me, I'll add it in later on. So here's a question that's come to mind. You've got all these agents, you're doing all these things, um, how are you keeping your token cost cost low? How are you making sure that at the end of the month you've not got a huge bill coming from using these agents?

Yeah, that was a learning curve, Mark. So I started with an API from Entropic, Opus 4.6, which was somewhat the recommendation for your chief of staff. Which oversees security and avoids any kind of prompt injection. So go with a top-of-the-line model. That was the idea. And I want to say, I probably burned through with very little at this point in time, like when I started, there was not that much usage. And I want to say I probably burned through $25, $30 a day, which for my taste was a little bit too high. So then I started to, you know, do some research, and what people basically recommend is that they say, okay, you want to have a top-of-the-line model for your chief of staff or someone to oversee security. Your sub-agents, you know, like Tim or Jeff, they can run on a cheaper model, and I chose GemiK 2.5, which is very affordable. It's through an API as well. And then I'm not even sure whether I should mention that because I'm not sure how long Anthropic is even allowing that to do. But, you know, for your audience, Mark, I'm more than willing to share that. So I changed my API to a token. So I have a Mac subscription for Anthropic. So it's a flat rate. And since then, I am not having any additional API costs. I'm just paying those $125 a month for the Entropic flat rate. And then again, Kimmichay 2.5, it's minimal.

It's less than a dollar a day. Wow. And you've got two really good use cases there. Before we go into like, uh, the final sort of questions, is there any other use cases that you've got where it's just been wow? And I really like how with yourself everything is evolving around the short-term rental industry because there's so many different ways we could go with this, but you're like really hammering it home to how a host could be watching this and getting started. Is there any other use cases that you've seen in, in your time doing it where like, wow, this is great?

Yeah, I have many more, but you know, I have to be very selective because we're running out of time, right? But what I really like is I have never been really a social media geek, not a fan of Meta Business Suite UI. So I have given this, delegated this task completely to Jeff. And Like only yesterday I said, hey, Jeff, there is like the 175th anniversary in Cape May, New Jersey this year. And he started to create completely on his own. Not only did he write those 15 social media posts for both Facebook and Instagram, but he also created images. So I have him connected to Nanobanana, so he can create his own AI images for anything he cannot, like, find copyright-free on the internet. And what he does, and this is like something which is very interesting, unlike ChatGPT or any other LLMs, you can have ChatGPT write up an article as No problem. What Jeff does is he does a whole series of posts and also creates those drafts directly on Instagram, on Facebook. I have not yet given him permission to post, so they are all saved as drafts.

Yeah.

He's identifying the right timing. So some of those events which are tied to this anniversary have not yet been announced. So there's a cron job. So every day he goes and scrapes certain websites to see whether those dates, those events have already been published. And then based on this information, he changes the schedule. So he basically says, okay, this post we rather post, you know, 3 weeks ahead of the event and not now. And this is all stuff, in my opinion, which those frontier labs cannot provide to you at this point in time. But it's getting to a point where there is also recursive learning. So the idea would really be that Jeff sees what impact he generates with a certain certain post and then learns from it. And then with his next post, he's getting a little bit smarter because the whole idea is what you have been preaching, Mark, get as much traffic to this neatly made website Woosley has helped me with.

Yeah, it's fascinating that you— and so you haven't yet given it the permission to post, but I assume in the coming weeks or months, you'll have the confidence to actually just even let it do that as well, right? So you'll just have a fine-tuned machine in the background that's learning all the time. I think that's the idea. Yeah. Wow. I really, really like that. Just from a point of view, from everything that I teach and talk about a lot, I say about posting on social media, it's so important. And every single time I get the response back, I haven't got the time, I haven't got the skill set. Now it's just it's going to be a case where you're right, where more frontier models can catch up and get really, really good, and they're giving the ability to anybody to do this. That won't be a reason in the future, which is exciting and scary at the same time. It is. Yeah. I made a prediction last year that I feel like it'll come a point in time where any digital message that you receive, whether it's a text or an email or a social post, you won't know if it's a human who's wrote it or a computer. And the more I see this technology ramp up, the more that I can just see this happening sooner rather than later. Um, okay, I would like one more use case, and then I'm going to hit you with the question that the previous guest left for, for your good self, and then we'll, we'll wrap this up. This has been fascinating. The, the, the, the last one, the one that you can go, this is a good use case, because because every single one of these— again, what I love, the guests today have been a mix of vendors, friends of mine from the industry, PMS owners, marketing agency owners, chatbot owners— to, um, to, you know, they've talked about a few use cases in property management, but yours is fantastic. Every single one, anybody watching can go, ah, this is great, I can use something like this. So what's this final use case that you've, um, you've got for us?

Well, this is maybe then not one which is directly linked to STR, but I think there are some learnings which people can extract from it. And I actually did that already. So what I've tried to do is to give one agent, JP, access to paper trading via Alpaca. He does day trading on my behalf. So far, it's only paper trading. What I'm trying— the experiment is basically not necessarily to, you know, really have him earning money. The experiment is really recursive learning. So I have tasked JP in the beginning to read everything up on day trading. Um, watch YouTube videos, identify the relevant books, um, which are helpful for him to perform this task. So he created a curriculum completely on his own based on this one prompt, and then in like nightly sessions, right, so a couple of hours per night, going through this curriculum one by one I always asked to get a morning briefing about what he actually learned. And then most importantly, apply this learning to the paper trading. And again, recursive learning, what every afternoon after the stock market is closed, what have you learned? What could you do better? What are your wins? What are your losses? And this principle, I actually already applied to to this STR agent, Jeff, where I said, dude, you have to start learn the STR business. You have to every night create a curriculum. Every night you go on, you know, relevant Reddit threads, you go on YouTube, you read on my specific market. And that is what he does now. So yeah, I mean, again, the paper trading thing is not that relevant as I think the idea of asking the agent to create its own curriculum and make your agent a highly specialized expert in the market you are operating in. Because that's again something which if you like go to ChatGPT you will get some very good but still very generic answers. This agent now is becoming— Cape May is a very small market. It's a very attractive, but it's a very small market. I think we are talking 700 listings or so in a very small geography. So very dense, but he is becoming a like kick-ass expert for this little niche STR market. And this idea you can apply to like any topic, any topic, whether you are an STR operator and you're like, you know, having, you know, you're catering to a market in Florida or wherever it is. I think this is the very powerful lesson.

Lesson from this, this last use case. That is fantastic. I wonder if Jeff has listened to any Bruce Lee podcast or read the transcripts of any Bruce Lee podcast. That's funny. Ah, you're right though, train them to be super specific on your niche and on your industry and on your market, because there's no point having generic advice if it's not going to work for a very hyper-niche location like the one that you mentioned there. Fascinating. I could talk to you all day. The trading thing fascinates me. So what is paper trading? Just for people like me who don't understand what the word paper trading means.

It's not real money at this point in time.

All right. Not real money.

So it's just literally money.

Yeah. It's not real money. Are you tempted to give it an account with like crypto or something like that and see what it can do? I probably will.

Wow. Just because I'm just too curious. I'm just too curious to see what he would do. I'm taking it deliberately slow, mostly for security reasons. But this entire thing is too fascinating.

And yeah. You are a phenomenal human being. I could talk to you all day. It's definitely fascinated me, and I will make sure that everybody— when you send me the, the X handle, I'll drop it in the comments later on. Um, so I've got a question from the, the guest before, but I feel like the guest prior has got a, a just as good question, so I'll give them both to you if you don't mind. So the question was, um, how are you implementing security security in your setup so that, like you say, you are very risk-averse, but what, what sort of security guardrails are you putting in place so that your agents don't just go rogue and, and take over? Like, what, how are you setting this up to where you are confident that everything that you have created has been done in a way that can't go crazy and spend all this money on your behalf?

Again, a non-technical person, I won't be able to give you the terminal commands because that all has been written up by AI. So what I basically did is, there were, particularly in the beginning, there were some almost do-it-yourself tumor-like YouTube videos where people said, "Oh, don't install OpenClaw." Those are the threats. So I asked AI, "Go watch this YouTube, get the transcript, identify those potential threats, and then guide me through how I can avoid those pitfalls." And I triple-checked that, right? So I didn't rely only on my OpenClaw agents to do that, but I triple-checked that in this particular case with Opus 4.6 and also Gemini. So I had like 3 different opinions on that same topic, let's say prompt injection. What is necessary in order for us to avoid any prompt injection? And you can just ask that AI, right? What is necessary? And then they come up with, I don't know which commands I ended up putting into OpenClaw or the terminal to ensure that this is not going to happen, but I have a high degree of confidence that because I saw the pattern, right? I do understand that if 3 models are basically saying the same, that there is something to it. And that is what I started to implement.

Does this make sense? Yeah, it does. Ask AI, it will guide you. But don't— I think the message I'm taking is don't trust it 100%. Use it to get you 80% of the way there, and then use 20% of your own human intuition to get you the final word away there.

So yeah, and what I always do is, Mark, is like whatever topic it is, um, I always ask multiple models and see what— if there's a major deviation, I get suspicious.

Um, very smart.

But, but I— yeah, very smart. Okay, so the question, the, the true question You said that there's a lot of— there can be a lot of doom and gloom with everything AI and AI agents. You know, you can go on YouTube, you can go in the media, it's not hard to find it. But the question that the guests wanted to give to you are, what do you see are the benefits for humans with all of this technology that is coming very, very fast? Like, what do you— if you were to like put the, the rose-tinted glasses on and look to the future and go, this is the benefit limit for humans to have this technology at the rate that we're having it, what would that be?

I think we are entering a time where if you are really a creator and not only a consumer, sky's the limit. I think it's very hard to tell where this all leads us to as a whole humanity. And what it's going to do with like, you know, unemployment and whatnot. But as you said, looking at it from an opportunity standpoint, this is the time for being an entrepreneur. I mean, we were in our chat group exchanging all those little apps we created, right? I created an app from an idea I had with a CrossFit buddy yesterday morning, and by the afternoon, I had a fully blown application with an LLM built in, in this case, Qwen from Alibaba. So intelligence included. Well, so long story short, I think I think the opportunities are unlimited. I am super excited. I think it's a very, very exciting time we are living in and something we would probably look at— and I think in particular this year and maybe next year, we are going to look back 100 years from now as a turning point in history.

Wow, what an amazing place to end. Hanno, thank you so much coming on. Uh, really, really appreciate it, and hopefully we'll get you back on another time. Any final parting words, anything you want to say to the audience before I let you go?

Oh, it was fun talking to you.

Thanks for having me, Mark. Much appreciated. Massive. Thank you very much. Okay everybody, that was it. That was the whole, uh, live stream. We've been going for a few hours now. We've had some amazing guests. Can you all please drop into the comments and just say thank you to everybody. And if you have enjoyed this and if you want us to do another one, then just let us know. If you leave a comment, let us know your thoughts. Let us know your OpenClaw agent. Let us know what you're using, whether it's Manus, is it Claude, is it OpenAI, is it Groq, is it all of these that are out there, or are you dabbling with the OpenClaw? Let us know in the comments. Thank you so much for tuning in.