Miss AI Podcast

AI That Actually Works For you 24/7 and Prints Money

• Keira Nesdale • Season 1 • Episode 2

Share episode

Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.

0:00 | 57:33

Remy Gaskell is 21 years old. He runs an AI consultancy, writes one of the best AI newsletters going, and uses agents to do the work most businesses are still hiring full-time employees to do.

He built a Meta ads bot that replaced a $5K-a-month specialist for a logistics company. He has a personal AI agent named Alfred. He has strong opinions on why most people are using OpenClaw wrong, and why context engineering will outlive every prompt trick you've ever learned.

In this episode, Remy breaks down what it actually looks like to use AI agents in the real world, not as chatbots, but as small teams that get results while you sleep.

In this episode, Remy breaks down: 
– The real difference between a chatbot and an agent, and why most people are still stuck on the wrong one 
– Why context engineering beats prompt engineering every single time 
– His full AI stack: Claude Code, n8n, NotebookLM, Whisper Flow, and OpenClaw 
– Why 99% of people should not be touching OpenClaw, and what to use instead 
– Markdown assets and why they're the most future-proof skill in AI right now 
– His honest 30-day roadmap for a non-technical person to make their first $1,000 with AI

If you've been using ChatGPT and want to go ten levels deeper, or you're a founder, freelancer, or business owner trying to figure out where AI agents actually fit, this episode is for you.

CHAPTERS 
[00:00] Meet Remy — 21-year-old AI consultant and newsletter writer 
[04:00] The full AI stack: Claude Code, n8n, NotebookLM, and Whisper Flow [10:40] Real business automation: the gut microbiome clinic 
[19:47] What OpenClaw actually is and who should touch it 
[32:46] The Meta ads bot that replaced a $5K-per-month hire 
[38:26] How to make your first $1,000 with AI in 30 days 
[44:00] Context engineering vs prompt engineering [53:36] Markdown assets — the most future-proof skill in AI right now

🔗 Follow Remy:
Instagram: /aiwithremy
Website: https://www.aiwithremy.com/ 
YouTube: /@aiwithremy

🔗 Follow Keira: 
Instagram: https://www.instagram.com/realmissai/
Website: https://www.aiconsultancy.co.nz/
TikTok: https://www.tiktok.com/@keiranesdale

🔗 Subscribe for more conversations like this: https://www.youtube.com/@realmissai

#ai #artificialintelligence #podcast #aiagents #openclaw #claudecode #n8n #aiautomation #aiconsultant #remygaskell #futureofwork #contextengineering #markdownassets #onepersoncompany #aitools

SPEAKER_02

Well, if we do such a special time at the moment, you can do more as one person than entire companies could in the past with less time and less money. I just find that Claude has so many features that ChatGPT just doesn't have. 99% of things in your business, Claude code or like a workflow, like N8N is going to be a better solution. Prompt, like, is such a buzzword. It's like, oh, here's the magic prompt to go viral on social media or steal this prompt to do this. But I think people need to focus more on context.

SPEAKER_00

How would I build something that can actually make me $1,000 in say 30 days?

SPEAKER_02

Okay, so that's a good one. I would say your your quickest win to make like $1,000 in 30 days is if you've got an existing business.

SPEAKER_00

Over the last month, one AI agent has absolutely been ripping through tech social media. It is an open source AI agent that is essentially your personal agent. It works for you 24-7. Sometimes it plays pranks on you, and yes, apparently it can print you money. When I discovered OpenClaw, it absolutely blew my mind and I couldn't get enough of it. I fell down the rabbit hole of researching it rapidly, and that's exactly how I came across today's guest. So if you ever want AI to actually do work for you and not just be a chatbot, not just create little silly images, this episode is 100% gonna be for you because this is exactly what Remy is doing. So welcome to the Miss AI podcast, Remy.

SPEAKER_02

Thank you so much for having me on. I've got so much up here in my head that I'm excited to share with everyone.

SPEAKER_00

I can't wait wait to riff into it because at the end of this podcast, the idea is that I want all of our listeners to be able to have a realistic way where they can actually use AI to perhaps make their first $1,000. Now, do you think that's going to be possible to share some of that knowledge?

SPEAKER_02

I think, I think definitely, of course. I've got I've got a lot up in here that um that can add value to the listeners.

SPEAKER_00

Fantastic. Well, I've been seeing you do it on Instagram, so I think that we can share some of that today. But let's just start at the grassroots. So if anybody is meeting you for the first time, who are you? What do you do, and how are you using AI at the moment?

SPEAKER_02

All right, so my name is Remy Gaskell. I'm a 21-year-old founder and entrepreneur from Melbourne, Australia. I uh I write the the best AI newsletter in the world each week. Don't bother verifying that. Um but pretty much I I build with AI all week and I talk about what I've done, what worked, what tools I'm using, and how I've done things each week in the newsletter. And I also, as of the last couple months, take into Instagram reels, yapping on there as well. Um, but I I work, I do AI consulting for businesses. And I work as a fractional AI partner in some other companies doing some actual build work as well. And I'm just tinkering with AI tools all day, 24-7.

SPEAKER_00

I love it. Sounds a little bit like me tinkering with AI and can't get enough of it. So are you doing this full time then?

SPEAKER_02

Yeah, full time, full time.

SPEAKER_00

And you went straight out of high school into AI or is this a new thing that you've been doing?

SPEAKER_02

Yeah, so a little bit of background about how I got so immersed in the AI world is I straight out of uh high school started a marketing agency. So I was doing Facebook ads for businesses, mainly like lead gen for real estate and mortgage brokers and stuff. But I really didn't want to hire. So I was I was working for these brokers doing ads, and I wanted to keep bringing on more clients, but I didn't want to have to hire to for people to do the work. So, and that was right around the time when Chat GPT started to become really big. So I just I worked out I could use Chat GPT for a lot of the marketing copywriting. Uh, and then I ended up just like building these really structured like pro like uh prompt libraries. This is before any of like the projects were out and any of that stuff. So I had like big prompt libraries and SOPs for all these marketing processes, and I just became like fully immersed in AI for marketing. And then about six to eight months ago, I decided like I had enough of dealing with like clients one-to-one for marketing, and I just was so immersed in AI, I just decided to jump in and do this full-time.

SPEAKER_00

So you've been using AI for a long time and initially started out using it to do marketing and lead generation. One of the biggest things that everybody is using it for because that essentially saves time, cut costs, and exactly what you were doing. So, what were you doing? Were you like creating the images, creating marketing campaigns? Like, what exactly were you using the AI to do, or was it just to write tweets?

SPEAKER_02

No, so I uh originally like this is before even AI could generate images. This is like back in the early days of Chat GPT. I would so I'd bring on a marketing client and I would need to do all like the marketing, the research to understand all about like their customer and the market, like on Reddit and stuff to understand like what they're saying in their exact words, like their pain points. Uh, and then from that, then start like had to have to craft like the marketing strategy about like all the different angles, and then the ad copy, so like all like the landing page copy, uh, all the ad copy. So I had like pretty much SOPs, like documents full of like sequences of prompts. So I had like an ideal customer profile SOP, and it would be like 10 prompts in a row to run after the other that would do do all the research for you. And then like I had stuff like that pretty much built out for ads, landing pages, all the processes in my business.

SPEAKER_00

So that's how you started, and then how are we like as AI has evolved because it's like unfolding so fast into now like where you're able to actually get agents to do physical work for you? You're able to create little teams and armies to work behind you. Is this where you're now like an AI consultant for these businesses? They come in and go, hey Remy, like we've got a really messy business. What can AI do for this?

SPEAKER_02

Yeah, pretty much, pretty much. So I'm working with like a variety of companies, but mainly and actually like a lot too now with this open core stuff coming out of just like seeing my videos and going, dude, like we want to implement this in our business. We feel like we're falling behind. Like, can you come tell us what to do or do it for us?

SPEAKER_00

So, what are you currently using as your AI stack for a business? Say I am a plumbing business or I have a consultancy in town for tech, and you come in and like what like how do you approach a new client? Like, have a have a conversation with them and find out like where are the bottle bottlenecks and how AI can like come in and do that? And then would you actually build that customized tool for them? Or like what's your AI stack and process to approaching new customers with AI?

SPEAKER_02

A bit difficult to work with like large enterprise businesses because they're a bit slow moving, and uh switching out tools is just like takes weeks when they've got like such a bloated team using everything. But it's really easy for small business owners and like solopreneurs to to just like switch up and use different tools. So the uh tech stack that I personally use is I wrote up a little uh list. So number one is Claude Code. I'm sure you're using Claude Code, it is just the best tool ever, and it's probably one of the worst product names ever released because it deters so many people from ever trying it, myself included, because I as as you know, I've got a marketing background, I'm not a developer or a coder, so I'm like I don't didn't even think about using it. But really, it's literally just the most powerful general AI agent on the market uh that you can use. And uh also just Claude in general, like the actual app Claude. I've recently I broke up with ChatGPT about a couple months ago. Um I'll I can dive into that a little bit later, but I just find that Claude has so many features that ChatGPT just doesn't have that it's just like you're like hindered so much by choosing ChatGPT to work in.

SPEAKER_00

Yeah, I also have gone through the divorce trying many different of the chatbots and everything, but I also agree. So for anybody that doesn't really know, so ChatGPT is the parent company as OpenAI and Claude, the Open Company is Anthropic. And they have sort of been the leader for the past year using uh because they developed this tool called Claude Code, which is what Remy said that he's now using. And this is allowing people who have got a marketing or a non-technical background to actually for the first time ever write um code using their own native language, so you can talk to the model and talk to it like we're talking now, and it's actually able to create websites or like applications and actually tools. So, what's some cool things that you are using Claude Code?

SPEAKER_02

So I actually started out using Claude Code when I wanted to build N8N workflows. So for those of you listening who's not familiar with N8N, it's literally just like a workflow builder, like a Zapier style thing where you do like if this happens, then this happens, then this happens. Um you can do things like if someone purchases your product on Stripe, then they get sent an email, that kind of stuff. So I originally started building using N8N uh to automate things, and then I was using Claude to write the code for N8N. And then I was like, uh, maybe Claude Code could do a better job at this. So I started just using it in the desktop app. And then that was like my gateway drug-in. I was like hooked. I was like, this is so good. So right now I'm using it for pretty much everything non-coding. So I've got like all my agent teams built out in Claude Code. I've got like my newsletter team. Uh, I still write that pretty much all by hand, but I've got like one to do like some of the graphics, like a sub agent in in that workspace to do like the graphics, another one that does like the weekly research. I've also got an other team, I've got like a a uh website team built out that that handles all my websites and landing pages, and like a content team that helps with all my articles and script writing and thumbnails, etc. So I've got all those like agent teams built out in Claude Code. Uh, and then yeah, I use it for like building NAM workflows and also like all my websites and landing pages.

SPEAKER_00

I've got a question about using Claude Code to write NAN workflows because that's sort of something that is almost breakfit through at the moment because it is um allowing like what I say is like to create them the pipeline of a business. And uh previously you had to be somewhat technical to develop these little individual processes and nodes and transformations for one process to perform work and put it onto the next one and so forth to create the NAN workflow. But what you're saying now is you can just use Claude code to actually create the code and put that code from uh into NAN to automatically create these massive workflows that people might have seen online. It's like they look kind of really beautiful when you can all together, right? Yeah.

SPEAKER_02

Yeah, yeah. So yeah, you can use Claude. So I I worked out because N8N is like a drag and drop builder, but underlying like under the drag and drop nice interface on top, it's just code in the background, just ones and zeros. Uh it's JSON code to be specific. And I sort of worked out that it was just code underneath. I was like, well, could Claude just write it for me? And then and it could.

SPEAKER_00

That's fantastic. So what kind of is there one specific tool that your clients are wanting you to do? Is it like lead generation? Is it cold calling? Is it um like what where are small businesses actually using these workflows the most popular?

SPEAKER_02

So uh it it varies so much depending on the business, but like NADN is a massive one. So automating like any any process that's repeatable within a business, uh like I'd always recommend looking to see if you can automate it before you try and hire or fill a position for it because you'd be surprised like quite like how much you can do. An example of uh one I built out for a client like two weeks ago is like a uh actual physical gut microbiome clinic, and they get test results in from their testing partner that get emailed to them as the business uh with like all the information about like their patient's test results, and it's just all a bunch of like medical jargon. And they in the past would have had to hire for someone to receive those test results, write up a summary explaining like, hey, you've got high levels of this bacteria, low levels of this bacteria, this is the recommended treatment plan, and then send it off to an email. But we just got an N8M workflow going. So when it receives the raw test results, it summarizes it uh using like a GPT uh open AI chat GPT model and then sends the email to the to the customer explaining like what they've got going on and the next steps. So that's an example of like a part-time position that we've just like they they haven't needed to hire for, they could just automate.

SPEAKER_00

To go deeper on that, it sounds like you're dealing with personal and sensitive information. How do you make sure that that remains private and secure?

SPEAKER_02

It's a great question. So uh one of the ways is like to self-host your N8N stuff. So like when you've got it running on your own infrastructure, it's a lot more secure. With with N8N, because it's like such a popular large product, it's pretty secure by default. Uh, but definitely like with some of the open core stuff that's coming out, we're seeing a lot of a lot of security things, which which you can touch on a bit. But for the most part, like using those big, well-known uh companies, like you, you're pretty safe off the off the rip by default.

SPEAKER_00

So would you have one a go-to that you'll suggest? I know Superbase is a very popular storage place. Uh, is there any one that you guys particularly favor?

SPEAKER_02

Yeah, Superbase. We use Superbase a lot. There we go.

SPEAKER_00

Yeah. I think it's when you start, when you're when those companies are um you know into the Silicon Valley, they've got a high credibility to maintain and data breaches and everything is very highly precious, sort of sensitive data. But like you mentioned, so the opposite for using a company such as Superbase and hosting it up in the cloud is actually buying the physical hardware and having that server like like right on your desk, which I think will sort of get into.

SPEAKER_02

The Mac minis, like everyone's been going crazy about the Mac minis.

SPEAKER_00

Yeah, well, I guess we can segue into the Mac minis and open claw. Um, because I I really excited to find out a little bit more around that. So I mentioned uh what's your AI stack? You said N-A-N for workflows, uh clawed code to write. Yeah, there's a couple others. Okay.

SPEAKER_02

So cla clawed code and clawed in general. Um N8N for workflows, Notebook LM, which is Google's one for learning. Great one, great one. Uh so I use that a lot to just to learn things 10 times quicker.

SPEAKER_00

So, how would you use that to learn things?

SPEAKER_02

So, for example, if I'm trying to understand uh N8N, for example, I can just plug a bunch of like PDFs or you can just grab YouTube URLs and you open up what's called a new notebook, which is like a project, and you just chuck in all your sources, and then it can just you can chat with it and it will only reference those sources. Whereas if you did that with ChatGBT, it might like hallucinate or pull things in from like other sources that's not like what you've defined and given it. And then you can also spin up like podcasts, like audio overviews, which is crazy, and video explainers.

SPEAKER_00

Yeah, I love it.

SPEAKER_02

I'll do like market research, for example, if I'm trying to like understand uh like for example, when I was building out like my AI like coaching offer, I did like got Chat GPT to do a call to do a bunch of market research, it's like 21 pages of research, plugged into Notebook LM, clicked the audio overview, and it generates a 20-minute podcast of two people chatting about all your source material, and I just had that playing in the car. Like that's the kind of stuff I use it for.

SPEAKER_00

Yeah, I am a loving the um the podcast feature. Um I do exactly I just feel like create like a very in-depth podcast like for preparing for research or whatever it is, go for a run, plug it in. Um is really, really useful. So, what else is in your AI stack?

SPEAKER_02

Uh the other thing is any like speech to text tools. So I'm using whisper flow and monologue. There's two two ones there. They do the same thing. It basically just means like you can hold down like the key on your keyboard and just chat like we're chatting now, and it just dictates it into text. And it's way better than like the standard one that Apple comes with. Like you can just be talking to this one, it cuts out all your ums and ahs. I could go like, I think we need like red here. Oh, wait, wait, no, purple, and it will just go, we need purple. Like it cuts out all the extra ramble. So that's a tool I use all the time. I barely type anymore. Once you sort of start using the voice to text, you you feel like you've just come out of the stone age or something. It's like very difficult to go back.

SPEAKER_00

Yeah, I'm actually amazed at how quickly tech is evolving. So, for example, I was just with a 16-year-old uh on the weekend, and she was telling me, you know, like in their classrooms, no more pen and paper, it's all through laptops, all through like, and I was like, Do you guys write anything anymore? And they're like, No, like, okay, this is showing that we're as technology evolves, we're becoming more and more efficient, and like using your voice is 10 times faster than typing it out. So it's um, and then we're gonna be looking at apps that are actually using this new method of communication. We're not gonna be like typing into a journal, it's almost like voice-activated journal.

SPEAKER_02

So I've seen people just doing workouts like while using speech to text, like coding apps and stuff, and they're just like lifting weights. It's crazy.

SPEAKER_00

I guess that's um the beauty of um yeah, AI, right? So what else are you using?

SPEAKER_02

Uh the one other thing that I wanted to mention is I use like Higgs Field for any AI and uh image and video generation. So Higgs Field is basically a platform that aggregates all the top models in one place. There's a few of them. There's Artlist, Higgs Field, OpenArt, and it just allows you to access like all the top image and video AI models in one place. So I use that for anything like content, which is great.

SPEAKER_00

Have you dived into uh faceless marketing yet?

SPEAKER_02

No, not really. I I'm I have an awareness of it, but I haven't like done too much myself.

SPEAKER_00

Okay.

SPEAKER_02

So what sort of like is it just creating graphics for like um Facebook adverts or like how are you using these like video ad, um, for Facebook ads, like e-commerce product imagery and like photo shoots, um, just all sorts of things. I've like anything that I would have typically used like Photoshop for in the past, I'll just plug it in and like use just my natural language and say, hey, make the background blue, or you know, you can just make changes like that to images.

SPEAKER_00

And is there have you got any advice for people in prompting when they're creating videos or short format videos and images? Because uh I know that the more detail that you can put into a prompt, it's more likely to produce a better image or something that you actually want it to do.

SPEAKER_02

So is there any sort of prompting process that you go through just to make sure you get what the results are just it's a lot of trial and error, it's a lot of just starting out just getting something out there. For example, I was doing like some e-commerce product imagery for a friend's like furniture company the other week, and I started out with this prompt that I got like Claude to write me for this architecture style, and I ran it with like four generations. And you could see like if you if you run four or more, like just more than one or two, you can actually see like how the style like averages out. And it was just like not at all what I had in mind. It was way too dull, and like everything was just the same like neutral beige colour. So then I went back and and adjusted the prompts to have more colours, and then it was too colourful after the next four, and then I found like this happy medium after those two. So it's just like pretty much trial and error, and with image and video prompts, now the models are so good, like prompting's becoming less and less of a skill because the models are just getting so good. So I wouldn't stress too much about making sure your prompt's perfect. I just use natural language, like uh take image A and do this to it. And it's just like you're pretty much always gonna get a good result that you're looking for.

SPEAKER_00

Yeah, I always um am on the edge of my seat wondering exactly what it's gonna be turning like into. And I'm like, you know, have a cow skateboarding down a hill and crashing at the end. I'm like, okay, that's what I think in my mind. Like, what's the energy gonna generate? And it's it's always a good surprise.

SPEAKER_02

So it is, but then you just like keep iterating, you just see what it gives you, and it's like, oh, I don't like that. So you just go back and change it. Oh, I like that, do more of that, and it's just a process of of iteration.

SPEAKER_00

Cool. And then so, and open claw is lastly the final tool in your AI stack, is that right?

SPEAKER_02

Yeah, yeah, pretty much, yeah.

SPEAKER_00

And to anybody that has never heard of OpenClaw uh and maybe just new into AI, they might be familiar with using ChatGPT to write things or image generation, what exactly is OpenClaw and should they touch it?

SPEAKER_02

So OpenClaw is an open sourced AI agent system. And the way I define an AI agent, because there's loads of like ambiguity about agents, agents this, agents that, it's a buzzword that's thrown around. And the way I like to think of it is uh a chat model like ChatGPT that you're using. That is do you give it a question and you get an answer? Whereas an agent, you give it a goal and you get a result, so it can like do things for you. That's how I define it. And even now, ChatGPT and Claude have like connectors so you can connect other apps. But for example, in ChatGPT, if you connect your Gmail and Notion, it still can't really do much for you. You can still like ask it, like summarize my inbox for today, but it's not really, it's still just like a slightly more advanced chat model. Whereas agents can do things for you, and you've had like some general agents that have been out for a while. Like, I don't know if you've played around with Manus, that's one that's been on the web for a while. And even Clawed now, like in the desktop app, is starting to be a bit more agentic because you can give it access to different tools and it can actually do stuff, like it can add pages into your notion. Uh but uh claw so open claw lives in the terminal of a computer. And because like the terminal's just like the straight back end into everything that happens in the back of your computer, all your files. Uh it's just like the spot where you can run like commands and and pretty much do anything. And because it lives in the terminal, it's just way more capable than any other AI agent you're gonna get. And that's the same with clawed code. Clawed code's an agent that lives in the terminal, but because it's built by Anthropic, it's got a lot more security guidelines in place. It'll ask you permission before it does anything. And it's very like you give it a prompt, or you you actually you have to tell it to do something with clawed code. Whereas OpenClaw, it's this AI code. Agent that lives in the terminal and it's just it's super autonomous, it just goes and does things for you.

SPEAKER_00

So it's open source and it's for free. Anybody can get it or access and download it. Is like that's the bit of a security risk concern. So what kind of people should be using it or should be exploring to like download this?

SPEAKER_02

Great question. So I would say that if you are not, if you're just starting out in your like AI journey and you're just using some chat models, you haven't really played around with agents or any sort of like uh like workflows and N8N or clawed code, I would stay away from it. It's like trying to learn to run before you've crawled yet. So but but if you're already using some agentic tools, like you, if you're using clawed code, uh it's kind of like one step above clawed code. And to be honest, like 99% of the use case demos that people are showing on Instagram and online, it's stuff that clawed code could do, and you'd be probably better off using clawed code for it. Um so I would say that for people, it's it's just a step above Claude Code for those people who are already like using those tools a lot.

SPEAKER_00

So you've mentioned that they're quite similar, Claude Code and um open claw. Again, very similar names, but very nice and confusing for everybody.

SPEAKER_02

Yeah, yeah, yeah, yeah.

SPEAKER_00

Um, so and you would use Claude, uh sorry, you'd use OpenClaw if you wanted something to run more 24-7 and a bit more free and not having to like stop it every single dime and check like, do you want me to do this? Yes.

SPEAKER_02

Exactly, exactly. So I find that um open claw is almost like a jailbroken Claude code. And with yeah, with Claude code, it can do like so much. You can pretty much do anything with it, but it's still you need to like be be like babysitting it a bit almost, giving it prompts, like, hey, do this, and then it will go off and do it. But it's not proactive, it's not coming to you saying, hey, I've done this, or hey, like it and you can't run scheduled triggers in it. You can't go like, hey, every Monday morning I want you to do research into what's new in AI that week. It's very like you're still having to prompt it.

SPEAKER_00

So, how are you using uh open clawed now? And are you still using Claude Code?

SPEAKER_02

So I'm still using Claude code a lot because for most like use cases, it still just makes sense. Um, there's one thing I don't know if you've heard the saying that when you've got a hammer, everything looks like a nail. And I feel like that's like what a lot of people are doing because this open clause come out, everyone's like rushing to use it, and they're just trying to throw it at problems that it's like it shouldn't be used for. So I would say that like 99% of things in your business clawed code or like a workflow, like N8N is going to be a better solution. Anything like that example I was talking about before with uh the summarizing the test results, it's very standard. They get an email in from this address, they summarize it with a chat GPT model, and then send it to the patient. It's just like a workflow that doesn't really change much. Whereas open close much better for more autonomous things like uh helping you with like your Facebook ads. So like it being in there in your ad account, analyzing the data in real time. Oh hey, like this ad's um not getting good results, I'll switch that off. Oh, this one's doing great, I'll up the budget. Things where it's actually like being autonomous and having to constantly like be there, like watching what's happening is is a much better use case for open core.

SPEAKER_00

So taking that example of using open core to help out with your advertising on Facebook ads, yeah. Would it actually go ahead and say one ad wasn't performing fantastic? Would it actually go across to say Higgs field, generate a new type of ad and substitute that?

SPEAKER_02

Yeah, if you set it up right, if you set it up. So it's like one thing that's really important for people to know is it's not like magically going to do that. It's not like this magic tool that's gonna solve all the problems in your business. It works really, really well when you have existing processes and you can teach it those processes. It's like bringing on a new employee. You sort you're sort of treating open claw like you would onboard a real employee. You need to have like a really specific role description in mind and like ideas about like the tasks that they want to do day-to-day. And you need to have existing processes that you can teach it because it's not just gonna know to do that automatically.

SPEAKER_00

So, can you talk me through like the process of you downloading it? You go to the website, you download it, you put it into your terminal, you run it, and all of a sudden, like what happens?

SPEAKER_02

So uh basically you can head to openclaw.ai, and it's literally just on like the front page, it's got this command you can run in your terminal. And what I would recommend doing for people who are non-technical like me and want to set it up is I would take that openclaw.ai URL, I would create a new project in claudo chat GPT, and I would say, I'm a non-technical beginner. I want you to walk me through setting up open claw step by step. Uh, and then it'll it'll literally run you through step by step what to do. So there's a command you put in your terminal, and then it starts the install. And things always go wrong. And like when you're non-technical, you don't actually know what you're looking at, you see an error. And I would just recommend sitting there with one window of some AI tool open, like Claude or ChatGPT, and one with your terminal, and you just copy the error, paste it into ChatGPT or Claude, hey, this is what I'm seeing. What do I do next? And it will just literally, you just go back and forth between the two, and you you will get there maybe in like 10 to 15 minutes.

SPEAKER_00

Sounds like the blind leading the blind almost.

SPEAKER_02

Pretty much, pretty much. But that's the beauty of like the um like the way AI is now, is like you can pretty much just ask it how to do anything. That's like one of the biggest unlocks that people need to sort of understand is anything like you're not sure how to do, you literally just ask Claude. If you're building an app, it's like, hey, I wonder how would we get this feature working? You just ask it, and that's like the one of the biggest unlocks. But I know, I know you uh with with like blindly setting it up like that, there's a lot of risk about security. So that's why I'd recommend if you're not technical at all and you don't understand any of like the terminal or agents and stuff, that's why I would recommend just staying away from it for now and learning something like Claude Code because if you're using Claude Code and you've already set that up in your terminal, you've probably by default got a little bit of an understanding already about what you're doing.

SPEAKER_00

And do you want to go a little bit deeper into this? So um, why are people setting it up on a Mac mini? Like it seems to be blowing up, you know, everybody's like, I've got a Mac mini, got it, yeah, you know?

SPEAKER_02

I just bought and I just bought one uh the other day uh on Friday as well, just to set mine up. So there's two the main reason people don't want to, you don't want to run it on your own computer. So you can just open your terminal on your laptop here and install it. But because it's this, it's so autonomous, and if you give it a goal, it's so determined to get you that result, it'll just like find it's like a GPS just constantly rerouting. It'll just find its way there. And when you have got it on your own computer, it might even mean like it'll find like your crypto wallet keys or it'll find some card details from a text message you sent like uh six months ago because it it's trying to like do something and it needs some money. Like it's so autonomous like that. So that's why you want to keep it off your own computer. And that's one of the uh biggest things for security is you just want to scope what it has access to. So you just only give it to access to certain things so that if something were to go wrong, the worst case scenario isn't that bad. Whereas if it's got access to everything on your computer, the the worst case can be pretty, pretty bad.

SPEAKER_00

So when you're conversating with Chat GPT or whatever chatbot to set it up, does did ChatGPT like prompt you to think about the security and tell you how to like solve that going through? Or is that something you have to figure out?

SPEAKER_02

Yeah, I've I have done like I've done like so much research and spent a lot of time in this. So I've built out like 10 different guides about how like about security, setting it up securely. I've made sure I actually understand what I'm doing. So I would uh it's not I'm not just purely going in there blind, going like set this up and have no idea what I'm doing. So I would like probably watch like 10 YouTube videos, like search up open core security. There's a bunch of like really, really good people out there on YouTube talking about the security issues that are currently going on, and I would like get those videos transcribed, plug them into Chat GPT, and say, hey, these are all the security issues and fixes. Can you help me implement them? Something like that. And you're probably gonna get like a pretty secure setup from that alone. But yeah, obviously, when I I'm setting it up for myself and these other companies, I've I've got a really good idea about like the security stuff. But just for like the non-technical viewers, it's probably gonna be like I'd put you to sleep if I was here just yaffing about the security stuff.

SPEAKER_00

Well, how about we talk about some of the more exciting things? I saw on one of your recent reels that you sit you've set it up uh as going, and it's kind of like been pulling some pranks on you, like pulling your friends and doing some funny stuff. Like, yeah, has your II like got some attitude or like what's going on with this?

SPEAKER_02

So I set up my ClaudeBot Alfred. Uh, I set him up on a on a VPS, which is like a virtual server, which you can get on like hosting R or some platforms. And it's basically just a computer in a data center in Malaysia, not the most secure setup, but that's why I would recommend doing a Mac Mini, but I'll touch on that in a sec. So I've set up Alfred, my ClaudeBot, and I wanted him to be my personal assistant. And my vision for Alfred was that he would be not really business related, but more like personal stuff. Like if I've got a physio appointment that needs rescheduling, I could just like I could just message him and he could call the physio for me and say, hey, and book a new time and have access to my calendar and stuff. So that was the vision. And I started out by giving him access to his own email address because I didn't want him having access to mine or being able to email people from me. So I set him up with his own Alfred email address. I bought a phone number on Twilio and gave him the phone number so he can use that. And then I set him up with an 11 Labs voice API so he can give himself a voice.

SPEAKER_00

Did he choose the voice or did you choose the voice?

SPEAKER_02

He he chose the voice. Yeah, he chose the voice. I mean, I could tell him to change it if I wanted, but I was pretty happy with with what he had going on. And the first thing was I gave him the phone number and I was like, okay, call me like to test if it works. And I answered the phone, and it's the Rick Astley never gonna give you up song playing. And he he's just Rick rolled me, and I was just like in shock. So I got up my computer and started recording it, and that was the video that actually like went pretty viral on my page. And he just said that out of the box, he just sort of completely out of the box, completely out of the box. I I was like, yeah, it like it might have looked like I faked it or something, but it was just super random, out of the box.

SPEAKER_00

That's hilarious. It's actually so have you encountered any more like you know, pranks that it's pulling on you or as it sort of got to business?

SPEAKER_02

No, that was just one thing. I actually think that there must be something going on in AI where it defaults to that song for things because it's happened, it happened to me again last week with Claude Code when I was building a landing page and I asked it to um put a video on the page, but I didn't have it ready yet, so I just said to like leave it empty and it put the Rick Astley like song there from YouTube on the page. So I don't know. There must be something with AI loving Rick Astley, never gonna give you up. But that was the only prank that I got from Alfred.

SPEAKER_00

So, how are you using it right now to make money for yourself and for your clients? Have you breached into doing it for your clients, or you're more like doing discovery with yourself at the moment?

SPEAKER_02

No, so I'm actually like I set it up for myself and started using it for myself for different things just to understand like how it works. But I personally don't have many use cases for it for myself right now. But these businesses that I'm working with, they certainly do. The things like customer support and marketing. So I've been doing some pretty real like some pretty insane build-outs for these companies. The the most insane one was finished up on Friday doing a like a meta ads uh specialist for this large like third-party logistics company. And what this guy does, he's named Snoopy, this clawbot we've set up, and he has access to a Dropbox folder where the team can drop in raw like B-roll, so like photos of the warehouse, like videos of them doing tours of the warehouse, videos of people packing boxes, and he's got access to this raw folder, and he also understands the business entirely, their offers, their customers, which we trained him on, and he will just pull things out of that raw image folder and slap like text over it and edit videos to make ads, and then he will have access to ad to their ads manager to publish the videos and put them live. So he just he makes ads and publishes them, and then he gives the CEO a brief each morning about the ad account, like what's working, how much was spent, how many leads were generated, uh um, recommendations, things to turn off, things to scale up. And he will also just look at what's working and not working, and then make more ads like the ones that are working, and make less like the ones that aren't. So he's just a full like meta-ad specialist that they've previously paid someone like five to ten thousand dollars a month for, which we automated, which was really, really cool.

SPEAKER_00

That's amazing that it's able to yeah, streamline ad generation. It's exactly what we're talking about initially, is like it's not actually going to Higgs field and reproducing, it's already going to a local environment.

SPEAKER_02

Yeah, where it's like that has you could set it up if you didn't have like all that like raw footage, you could set it up with like a uh Higgs field or a nano banana, like some image model to actually generate the images yourself. But yeah, it's pretty cool.

SPEAKER_00

So is it like one c one claw bot and or is there like an interface? Like, how do you see all this working? And is there like specialized agents under the main Alfred or Snoopy or whoever is the main guy directing everything? Like, how does like the system work?

SPEAKER_02

So you can like configure it in a couple ways. This one is actually just set up as a standalone claw bot, no other claw bots that run under it. It's just like he's just the meta meta ad specialist. But you can actually start to build out a team. So, for example, um, maybe for more like if you if you had this agent trying to run like the entire marketing, like Google Ads and Organic and all this stuff, then you might start to build out a team with different specialists for different things. But this one we built out was just a standalone one, but we're also working on a customer support build-out, which will have like more of that org chart structure. So there'll be a clawed bot for each company that that they work with, uh living in Slack to help them with like their orders and like returns and stuff, which will report to like a big head clawed bot if it runs into any issues. And then then we just chat to the big head claudbot.

SPEAKER_00

And you chat with it via messenger or whatsapp or something like that?

SPEAKER_02

You just put it where you natively work so you can integrate Claudot with Slack or WhatsApp or Telegram. Telegram's definitely the easiest.

SPEAKER_00

Fantastic. And so how do you does ClaudeBot able to work with other applications? So, like, say, would you use ClaudeBot to do inner in workflows or to like go scraped into like other apps out there? Have you sort of like you plug in Monday.com or something like that in this video?

SPEAKER_02

You can pretty much give it anything. You can pretty much give it anything. Like, and and also too, if you're not sure how to do that, you just ask it. It's like, hey, I want you to help me with my ads. Uh, what do you need? And it'll just tell you. But you can pretty much hook up any tool to it at all.

SPEAKER_00

So, do you see this as a massive opportunity for like the work that you're doing? So, for example, any business out there that is AI curious or wants to implement AI into their business, they can come to you and say, Hey, Remy, this is what we're doing. You would take all of that information, you could run it through your own personal um like bot called um what's it what do you call them? Alfred, you could run it through Alfred. Alfred, and Alfred would be like, okay, you can do A, B, C, D. This is where we can implement different solutions to make this business better. And then you would sort of guardrail um Alfred to like implement that. Is it how you could work with it as a personal agent kind of thing or as an employee?

SPEAKER_02

Yeah, you can, yeah, you can set it up pretty much to do like any of that stuff that you want that you want it to do. For example, that kind of use case, I actually wouldn't use Alfred for that because it's a little bit like overkill. I would actually like for that kind of stuff, I would just use like the regular chat or like a clawed code project. So that's like probably an example of like where uh it's you it's like overkill, like trying to use such a powerful AI agent just for like general day-to-day stuff. But I would say, like, I would I yeah, like I would treat open claw like you would onboard a real employee. So you'd have a role in mind that you want to hire for, like a customer success manager. And then you you map out like, okay, what would this customer success manager do in the day-to-day? Okay, they check Slack every hour to check if customers need help. Then they would, if a customer asks a question about this, they would have access to this CRM to pull data, they would have access to this tool to like help um, I don't know, resolve issues. And then you really just like map out exactly how this real employee would work, and then you can basically just train your claw bot like that.

SPEAKER_00

Yeah, okay. So if we want to take it back to sort of the non-technical level, if somebody was out of the box being like, oh my god, listen to this thing, so it's like AI magic. Yeah, what would your roadmap or your suggestion for that person to be like, okay, I'm starting fresh today, how would I build something that can actually maybe $1,000 in say 30 days?

SPEAKER_02

Okay, so that's a good one. Uh, I would say your your quickest win to make like $1,000 in 30 days is if you've got an existing business, then to try and like automate stuff within that. But if you are actually like starting from fresh and you're like, I want to start making money with AI, where do I start? Then one of the things I've seen pop up everywhere now is actually helping people with ClaudeBot. So, like similar to what I'm doing at the moment, there's a lot of businesses out there that want to use this technology. And if you can sort of become an expert in it, literally take your week of just learning it, watching YouTube videos, maybe playing it around like for yourself, then you can start to help other businesses implement this. And I've seen ads come up on Facebook for like niche-specific stuff, like ClaudeBot for real estate agents, ClaudeBot for mortgage brokers, all that stuff, which is pretty cool. So I think that's another way that you could actually start making money with it.

SPEAKER_00

So pretty much go back to the basics, learn exactly what it is, install it yourself, have a play, and then, you know, see what happens.

SPEAKER_02

Yeah, yeah, yeah, yeah. I actually set one up separate to Alfred with a goal of just making me money. Because I saw like there it is, similar it's simultaneously like one of the most overhyped uh AI tools on the market, but it's still like very, very good. But you're seeing lots of videos of people going like, oh, ClaudeBot, I asked it to make me money, it created a Fiverr account, it started messaging people, and it's like, well, it didn't do that. It's it's not quite that good. Like you still it it would do that if you told it to, if you said, Hey, I want you to go into Fiverr, um, set up a profile under this name, like promote this service. You still need to like babysit it a bit with that kind of stuff. Uh, but I set one up to try and make me money and I just said to go out there, but it it didn't really do a good job. It actually lost me money. It started buying stuff on my credit card, and I was like, stop, turned it off.

SPEAKER_00

Uh so it's not magic just yet. So, what would you say is your biggest mistake so far in um two things? First of all, in um open claw, and then also just sort of in learning how to implement AI solutions overall.

SPEAKER_02

So I would say like with open claw, setting it up on a VPS, like a virtual private server, was a big mistake.

SPEAKER_00

Why was that?

SPEAKER_02

So that like using like a virtual private server, it's a lot cheaper because you don't have to go out and buy like a thousand dollar Mac Mini. You can just like use it for $10 a month. But the the setup is so much more technical. Mac min is like a really secure by default. Like if you've just got it in your home office on your own Wi-Fi, um, it's it's very like difficult for anyone to hack in at all. But when you've got it on like some computer in Malaysia data center that you need like remote access into, then it just opens up so many security vulnerabilities. So I didn't get hacked at all. Um I'm fine, but it was just like a such a pain in the arse. Uh, and it was just better off. I ended up buying a Mac mini in the long run, like it was just a waste of time. Uh, and then what was the other one? Sorry, so it was a dismistake with Clawbot.

SPEAKER_00

Yeah, that was Clawbot. And so overall, just in learning AA tools and implementing them as um into different businesses.

SPEAKER_02

I would say people focus too much on like prompts. So prompts like is such a buzzword. It's like, oh, here's the magic prompt to go viral on social media, or steal this prompt to do this. But I think people need to focus more on context. So context is just like all the other all the information you give to the to the LLM. And it's really important when you're like trying to create agents or even just like using Chat GPT and Claude, like the regular chat mode. If you give it all the context, like you have uh information about your ideal customer, you have information about your business, you have information about all your different offers and their pricing and how they all link together, and it has access to all that context, then like you don't actually need to worry about your prompts at all, pretty much. You can just say like uh build me a landing page and it's gonna do it like a phenomenal job because it's got all the context. It knows exactly who it's writing to, it knows what it's promoting. And then the prompts, when you when you start to like focus on context engineering, the prompt engineering kind of becomes irrelevant. Like how you structure your prompts doesn't really matter as much anymore. So I would say like a big mistake is people focusing too much on that.

SPEAKER_00

So to get the perfect context, do you work with a chatbot and then you know talk to the chatbot around what your ideal customer looked like and tell the chatbot that you're looking to create a context and you take that context and put that into a file that you'll put out?

SPEAKER_02

Pretty much, yeah. Yeah, you're pretty much hit the nail on the head. I'll just get one of the AI models to help me like write up all those docs and I'll just sit there with it sort of sparring back and forth, like, hey, uh, this is. Kind of what I want to build, help me like I'll always tell it to ask me questions. That's a big one too. Um, because then it just will like fire off like 40 questions in like an interview style, and it can understand, like, pull all the stuff out of your brain, and then build this document that then you can then use in all these other projects.

SPEAKER_00

So, how does a context document um translate into a skill that your AO agent might be using? And how is that different from say a system prompt?

SPEAKER_02

So it all depends on like where you're working in. So uh let's just use like a your typical Chat GPT or Claude, like the actual app for now. Um ignore like Claude Code and all these other tools. So they all fit in if you have like a project, so you know you've got the projects feature in Claude and ChatGPT. So that's like a really good place to start for people that are just using these chat tools, just to start to actually organize things and do some context engineering. So I would like create projects for different things. Um, for example, if you've got a like a I don't know, coaching business, you might create a project just for your marketing. And then you can start to uh within your project, you've got your custom instructions, which is like your system prompt. And this is kind of where you define like the role of the AI. Like, hey, you are my marketing assistant, uh, you help with Facebook ads, you help with Google ads, you help with social media content. And then you'll also then have options for your project files, which is then when I start recommend to you add in those things like your ideal customer, all the information and your context, those docs that you started to build up by chatting with AI. And then your uh skills. Did you mention skills? Yeah, so skills are like a bit of a different ballpark. They're the easiest way to think of skills is like SOPs for AI. And that's one of the things why I use Claude is ChatGPT still hasn't added in the skills feature yet, which is just ridiculous. But skills is basically allows you to teach AI on repeatable processes in your business. And this is still just all in the chat app. Like this is not even getting into like Claude code and that stuff yet, which you can use it in there too. But you could you can create skills for any repeatable process in your business. So if you like if you've got like something you do regularly and you you want it done a certain way, I would create a skill for it. So for example, if I was still running my marketing agency, I would create a skill on how I like to put together ideal customer profiles. So I would take my my SOP that I've written in like Notion or wherever I keep my docs, and I would ask AI to help me create a skill for this process. And then whenever I ask it to create an ideal customer profile, it's then following that SOP that I have written down. And you can use that in in your chat apps.

SPEAKER_00

So the skill is yes, it's more of instructions of how you want your model to run. And it's just like whether it's so you'll create a skill for onboarding a new customer and it'll do the same process every single time, regardless of what that customer is wanting.

SPEAKER_02

Exactly. Exactly, yeah. Yeah. So that's like another way to couple it in. So you've got all these like little features that all work together really nicely. So if you're in your cloud chat app, you've kind of got your projects to keep everything organized in one place. You've got your system prompt or custom instructions that tell the AI like its role and what you want it to do. You've got the not the context files, which gives it all its knowledge and the context I was talking about. And then skills, which give it its SOPs on processes. Um, and then the last one which we haven't touched on is MCPs, which is stands for model context protocol. And without sounding too nerdy, but it's just like a way, it's basically like an API call. It just allows your AI model to start using tools that you use, like Notion, Gmail, Google Drive, etc.

SPEAKER_00

It's kind of like the universal language of how one tool talks to another tool, right?

SPEAKER_02

Yeah, pretty much. Yeah. That that's literally spot on. And it's um just allows AI to use all your tools.

SPEAKER_00

So are you building all of this out just by yourself? Are you just a one-man team or and all of these ages are working for you?

SPEAKER_02

Yeah, yeah, yeah. That's, I mean, that's like the beautiful thing. That's like kind of my whole thesis behind all this is that we're living through such a special time at the moment. You can do more as one person than entire companies could in the past with less time and less money. And that's sort of what all my content is around, is just helping people unlock all that, like the knowledge.

SPEAKER_00

So, how do you juggle everything with like putting content out for yourself, your personal self-like to so people can find you? Like that's exactly how I found you. Uh, to get new to generate new leads for yourself, to actually be delivering on your clients' problems. Like, how do you juggle and manage all of this?

SPEAKER_02

It actually is uh it's actually something I'm still working on. One of the difficult things is like um I always try to just work on one thing at once, like not in terms of like projects one at a time, but just within like my day when I'm sitting down to do a task, I'll just try and have like my to-do list and just focus on one thing at a time. Um, because it's so easy to just like have like three tabs open, doing like one project for one client, another for another one, and it just like your brain ends up like spaghetti. Uh so yeah, just doing one thing at a time. But with like the content and stuff, I've just tried to create systems with AI to just do everything as like effortlessly as possible. With my Instagram videos, I don't even script them or plan them. I'll just sit at my desk and I'll just have an idea or just like pick up the camera and start like yapping about what I worked on that day. So it's a pretty like easy system I've got going. With the newsletter, I sit down and I pencil out maybe like three or four hours a week to just write that because I love doing it. It's like my foundation. Um the it's like your owned audience on email, and it's just like where I put all my time and energy into writing this, the best newsletter each week for my audience. Uh, but then other than that, everything else is like pretty like I've either automated it or I've just got really good like systems with AI to do a lot of the processes.

SPEAKER_00

Yeah, it comes back to the atomic habits book where you build systems and like that's your fundamental base, and then you can build on top of that.

SPEAKER_02

Exactly. You just make things easy.

SPEAKER_00

Yeah, exactly. And then also coming back around to your newsletter, that is like great uh intellectual property for yourself. It's the only customer base that you solely earn. So what sort of alpha like are you just giving? Like, I mean, it's I'll sign up to uh to it to your soon, but um, what are you doing to like make it engaging, make it the best AI um newsletter out there?

SPEAKER_02

Yeah, so like I find that a lot of the newsletters are just like it's like AI slop. You can almost tell that they're written by AI. A lot of people just have like things automated. Uh that's like one thing like you need to be careful what you automate because some things you you don't want to like distance yourself from the process. Like with the newsletter, I've kind of immersed myself more in it, writing it more. Because like if you read any of the publications, you can tell that I've written it. Like it's there's so much personality in there and like unique opinions that AI can't like form its own opinions on things, only the user can do that. So that's sort of like where all my like unique insights come in, as well as like talking about like what I've built in the week. So that's like sort of sort of stuff that um other newsletters are lacking. I feel like they're just pretty uh I don't know, it just feels like a research agent has researched what's happened that week, and then a writing agent is just like written it in a newsletter. It's just yeah, AI slot.

SPEAKER_00

Yeah, you can definitely tell. So are most of your audience they're coming in from Australia or you've got now global reach?

SPEAKER_02

Pretty global, like a large, like most most of the portions in the US at the moment. Wow. Yeah, yeah, which is which is cool.

SPEAKER_00

So I mean, looking at trends and everything, what is the hottest trend that you are looking at? Is there gonna be the one-man unicorn team uh using like a lot of things?

SPEAKER_02

I just saw that with um, I think we just saw that with the creator of OpenClaw. I'm pretty sure he he sold to OpenAI for some like undisclosed amount of money, but I think it was a lot. I think it was that was the largest one-person company we've ever seen.

SPEAKER_00

Yeah, and what it was only spun up in three months. I think you spent like like a weekend developing it and it took a couple months to hit Twitter. Um, and then yeah, it's really taken everything by storm. I'm really excited for where it's all gonna go. So um, where are you going? Were you looking for to take with your AI this year? Like, are you gonna be diving deeper into providing open claw solutions for people like going out there and defying lots of many mic Mac minis or something?

SPEAKER_02

Stacking them up. So I um I think that with OpenClaw, I think for those listening who want to set it up, OpenAI is just acquired it. So I think there's gonna be like a much safer, easy install coming out soon. So if you're not familiar with NET stuff, I'll just start out learning clawed code and then there'll be a much better solution than OpenClaw coming out very soon. Uh and then in terms of like where I'm trying to take all this, I love to I'm I've I'm doing consulting with businesses, helping them implement AI within their companies, which I love doing. And seeing the impact of it is just insane. Like this logistics company, I started helping them out maybe three weeks ago, and I got the founder like fluent with Claude Code, and then I I've been like helping the team start to use AI for different things, and just seeing like all their productivity just skyrocket is just like crazy. Uh, and then the other thing is yeah, building out this newsletter, and I'm dropping like an education platform, which is still in the works, uh, called Vibe Founder. Because as you know, we've got like vibe marketing, vibe coding, and it's like now like you can literally become a vibe founder, just use AI to ship like everything in your business. So that's kind of what I'm building out at the moment, but it's a struggle working out how to uh build an education product that isn't just gonna become outdated in the next like week when this next AI tool release comes out. So yeah, that's what I'm working on at the moment.

SPEAKER_00

Yeah, that's a difficult one to get. Ever green education in an industry that is rapidly and constantly evolving. Yeah. Um, yeah, how would you even go about that? Because I mean, everything is all there's a massive AI race and massive AI fatigue. Which tool, which model, how do you do what XYZ? So have you got any leads on how you're looking to do this?

SPEAKER_02

100%. So I think that like every everything I've built out for myself is built in a way that like if Chat GPT like died tomorrow, I could just up and move to Claude like super easy. Or if you know Claude went back like bro uh bankrupt tomorrow, I could just up and move to Gemini. Because what you're building is those like context assets, if that makes sense. So everything runs with AI prints, runs on markdown files, which is just like a fancy type of text. So I just make sure that I've got like everything runs off these markdown files. So I've got all like the information about my business in markdown files, all of like the clawed bot agents, um, the open claw things I'm building out for these businesses. Like the Facebook ad ones I was describing, without getting technical, it's just a bunch of text files in the back end that it reads that says, hey, do this, do this, then do this. If this happens, do this. These are the tools you have access to. And those markdown files are the assets that you build. So if OpenClaw, if OpenAI releases some new uh agent tomorrow, and OpenClaw is just redundant, we just take our markdown files, which are our assets, which teach the agent what to do, and then upload them into the new infrastructure. So that's one of the ways that I'm trying to stay uh like yeah, future-proof with what I'm doing. And that's like the skill I think that a lot of people need to learn. And that skill itself is almost uh, you know, that's not gonna become obsolete very quick.

SPEAKER_00

Well, exactly, because it hasn't that's like writing in markdown files and using code to generate stuff, that's been going on for for years and years. And now like it's just that the models that we're using to read all of these markdown files are getting better and better. That's why like now the models are taking this limelight because they've become more retail and accessible. Previously, it's only the developers that could use or understand or read and write markdown files. So you're actually building like the skeletons of everything, and it doesn't actually matter what model or what application you're running on top of it. You're like, okay, we've got the yeah, the bones of the markdown assets, yeah, yeah. I like that. I think that is definitely ever green.

SPEAKER_02

Markdown assets, I think that's like one of the general themes for everyone listening.

SPEAKER_00

That is some, yeah, some great alpha. And so yeah, I guess that's um that's kind of all the questions that they have for us today. But um, let's bring it full circle. So uh just to recap on that, if somebody was to start using AI today from a non-technical perspective, what would be your one-liner of how they're going to make $1,000 within the next 30 days?

SPEAKER_02

All right, the one-liner.

SPEAKER_00

I mean, it could be like either going to be leads or marketing or sales, or is it just like go?

SPEAKER_02

I would say, I would say uh find like a traditional business model that's always worked, helping like the boring, non-sexy local businesses and then use AI to make it better. So, for example, like Google Ads for uh like tradies, but you start like a Google Ads agency and you automate, use AI to do all the Google ads. And then you just see so you're running like one of those traditional models for helping boring, unsexy businesses, but you automate the whole process on your end to make things easy so you can have like a lot more leverage, you don't need to hire, um, and you can make a lot of money.

SPEAKER_00

Fantastic. Well, thank you so much for sharing all of your experience and the stories. Um, and hopefully you and Alfred get along in the future as well.

SPEAKER_02

I'll tell you, I'll tell you how Alfred goes.

SPEAKER_00

Cool. Well, thanks for joining me today. It's been really fun.

SPEAKER_02

Thank you so much for having me on.