I Built an AI Agent That Works While I Sleep (Cheap & Easy)

I'm over here still waiting for a good reason why you aren't building AI agents yet. And don't tell me it's too complicated, you don't know how to do it, because that's what this video is for. I'm going to break it down for you, super simple. And why? Because AI agents can run entire side hustles for you in your sleep. Lead generation, customer onboarding, content distribution, all of it. And you can do all this in one of two ways. You can learn what's called Raspberry PI, you can set up a little computer in your closet, and you can try to learn hardware and software at the same time. Don't do that. Or you can use a couple very simple tools, like one called Hostinger. I'm going to show you how to use this simple tool called hostinger to build a content multiplication agent from scratch. And I'm going to throw in a bunch of real business ideas that you can use today. So let's automate. First of all, let's talk about what an AI agent actually is and how it differs from an automation. So thanks to tools like make and Zapier, we've had automations for a decade now. And an automation is like an if this, then that machine. If I get an email flagged as important, then text me the subject line of that email. That's an example of a simple automation. No AI involved, no agents are involved. They just follow the rules exactly as you give them and they don't think, adapt, or even understand what's happening. But an AI agent is like a digital intern that understands, that can adapt, that can make decisions on the fly. It doesn't need to be told every single rule or every single step. It just goes, yes, it needs good prompting, it needs good direction, and it might break sometimes, but it's about four to five steps further than an automation can ever go. For example, an AI agent could check your inbox, read all the emails, figure out which emails are leads, which emails are spam, which emails are from your grandma, and decide what to do next. It can put the leads in your CRM. It could add a reminder to your reminders app to call your grandma. It be could, and it can schedule a meeting for you all without having to write a single line of code, or without you having to create a simple if this, then that automation rule. AI agents use language models like OpenAI, Groq, Claude, Perplexity, and then it has memory, so it can remember old conversations, and then it can also link up with other tools and goals so it can better plan and act on behalf of you. So an automation is like a fixed set of pipes. They push something from one end to the other. And an AI agent is like a flexible brain. It's ever learning and expanding. So we're going to talk specifically about two different tools and how they work together. Tool number one is called N8N. This is a tool that lets you automate tasks between apps with the help of AI. Kind of like Zapier, kind of like make, but more AI native and more complex. It enables you to do a lot more than with Zapier and Make, but it requires a lot more knowledge and training. But you have to host these agents and or automations on your own server, and that can be a little complex. And so the second tool over here is called Hostinger. Hostinger is a company that rents out these servers, so your apps, websites, AI agents, automations, integrations can all run on their servers. So you don't have to be technically advanced to know how to do this stuff. So N8N and Hostinger work hand in hand for the majority of people out there, like me, maybe like you, that don't know how to code, that don't want to spend seven hours figuring out how to set up a Raspberry PI. So with Hostinger, you can rent out a vps, a virtual private server, which is basically like your own little computer in the cloud, and all it takes is a few clicks. You take N8N, you install it on that server just as you install iOS20 on your latest iPhone. And then once it's installed, you can log into N8N through a web browser to create automations that run even while your laptop is closed, even while you're asleep. And remember, these agents, these automations, they can connect to YouTube, OpenAI, Instagram, any social platform, any API, basically any piece of software that has an API, which is most legitimate pieces of software. And I will warn you, this stuff is pretty addicting. It's pretty fun. Now, beware of what I call the cloud pricing trap. So N8N offers cloud hosting that starts at 20 bucks a month. But once you have half a dozen automations running, or even one automation that's just very active, those costs can quickly balloon way past a hundred dollars a month. It all depends on what your use case is. But N8N itself, not N8N Cloud, is free and it's open source, and so you save a ton of money if you can host it yourself. That way, it's the same price for 10 agents or 100 agents with, with unlimited usage. Kind of a game changer. All right, so let's go here and get some N8N hosting. We're going to go services self hosted N8N. We're going to choose the KVM2 plan. I'm going to go ahead and choose 12 months to lock in those savings. I'm going to make sure that N8N is checked right here. Go ahead and type in this discount code. Boom. Thank you very much. Submit payment. We're going to get started. Server location Phoenix. That's pretty close to me. When I get here, I'm going to type in N8N. Choose that, make a password here. Click finish setup. Now it is setting up, should take about five minutes. Okay, ready? All right, it's done. We're going to click Manage app set up owner account for N8N and we are ready to go. So if you look up here on the plus sign, here's where I'm going to build my workflows, AKA my automations or my agents. Here's where I'm going to add some credentials, some passwords, my API keys from other platforms when I'm trying to get stuff to talk to each other. And here's where I go edit my project where all these automations or agents will live. Now if I go into the personal section here and click on executions, this is basically showing me the completed tasks that my agents have already done. It's where I can see the results of my work. And then down here on the bottom left, I'm going to click on templates. That opens up this page right here where I can find hundreds if not thousands of pre built N8N templates for most anything I would ever need to do. Look at this. Auto create TikTok videos with V IO, AI avatars, 11 labs and GPT4 automate YouTube thumbnail creation and social publishing with templated IO. And I don't even know what some of these tools are, but you can see that it provides you with a ton of templates so you don't have to build them from scratch, which is awesome. But if we really want to learn, we should probably create some templates ourselves. So I'm going to show you exactly how to do that. I'm going to click on Create workflow. All right, so this is what I call my YouTube video to LinkedIn post generator. I'm this to run every day on a schedule and trigger to run at 6am every day. So I'm going to click on this Then I'm going to click on get many videos. I've already connected my YouTube credentials, so that's really easy. I don't need to show you that. I picked the option to get many videos, but I'm only going to have it return one. Now I'm going to put in the ID of my YouTube channel which I get from my settings. This is telling the node where to pull the videos from. Then I click Execute Step and it will spit this out. Right here on the right, you can see right there is the ID of the video. And that is the most important thing from this whole step. Right here I'm going to add the order, which is by date. So that's going to show me the most recent video, which is what I'm looking for. But remember, I'm not extracting the video, I want to extract this transcript from this video. So I'm going to click the plus here and then click HTTP Request Node. This will enable us to send a request to a website called YouTube Transcript IO. This is just a third party tool that will more easily enable us to get the transcript from the video. So I went ahead and set up an account and right here I'm going to send what's called a post request to this website and I'm going to check this box to send headers. I've got to generate my API key from this website, copy it. I'm going to call it authorization. I'm going to paste it in as the value and then under the body field I'm going to choose JSON. Then I'm going to type this in just as shown and I'm going to drag and drop the video ID into this part right here. You don't need to worry about how the code works. Just copy this transcript, save the work, and then click Execute. Snap. Oh, looks like it gave me a little error here. Yep, looks like I just have to put BASIC in there. Now if you look here on the right side, you can show that it's giving me the transcript, but it's all split up. I need to figure out how to combine all these. So I'm going to copy that and paste it up here above. And you'll see that it now combines the transcript successfully. Awesome. Now I can go ahead and delete this bottom half. I want to be careful to not edit out these hangups and these mistakes because it's an important part of the process. Now I'm going to click this button over here. Type in OpenAI message model. I already have my API Key in here. Then I'm going to add in a system prompt right there and I'm going to use GPT5. You are helpful, intelligent use YouTube transcript analyst. Here's the full prompt. Analyze this podcast transcript from the Kerner Office. Extract the three to five most valuable insights that could resonate on LinkedIn. Focus on actionable business advice, compelling stories, and quotable moments. Here's the transcript. Put that under prompt. We're going to drag that from the left down here. Now it's running right here. It's finding insights. And while it works, we're going to work on building the short, medium and long form LinkedIn post from this transcript from this output. Now, if I wanted this to be even more robust, I would add like example LinkedIn posts, more reference data to pull from. But for the purposes of this instruction, I'm going to skip that part. So I'm just going to name this one, find best points. Then I'm going to name this next one short post, then duplicate it. We're going to write the same system prompt. You are a helpful, intelligent LinkedIn post writer. And then this prompt will say, create a short LinkedIn post under 300 characters, punchy with strong hook from the Kerner Office podcast. And we'll use GPT5 for this again. So I'm going to put here in the prompt based on the insights given below from a recent Kerner Office podcast. Make it punchy and actionable. Use a strong hook that is under 120 characters and on two different lines. Do not use any emojis or flowery language. Write in a spartan tone of voice that helps these LLMs not just go off the rails. Right here under prompt. I'm going to write podcast insights. And then we're going to duplicate it. This will be the medium post. I'm going to duplicate it again. This will be the long post. And then I'm going to change the requirement over here just to say, medium length. Change the number of characters to 800 to 1200. And then on the long post here, I'm going to say create a long LinkedIn post, 1200-1500 characters. So now this is formatted to spit out three different results based on three different prompts, short, medium, and long. Each of these nodes are now generating a post. Oh, looks like it got hung up a little bit. I have to make sure that on this first of the last three steps, it's pulling from this node right here. So I'm just going to drag it from the left over here to the middle. And then on the long post node, we're going to do the same thing. We just have to make sure that each of these three nodes are pulling from the right place in the right order. Looking at the three LinkedIn posts that it generated. Pretty good. Pretty good. Yeah. I like all three of these. This is fully working and functional and ready for to use for yourself or to sell to others. So think of all the ways you can monetize this. If you're watching this, you probably don't have a podcast, and that's okay. I wouldn't have made this about a podcast automation if I didn't think you could profit from this somehow. You could charge between 300 and $800 a month per client. 10 clients equals call it $5,000 a month and that's almost pure profit. You could charge two to $5,000 one time for a setup fee and before you know it, you're at $20,000 a month in revenue. And this is one of thousands of templates, thousands of automations that you could build for people. Real estate, fitness, restaurants, healthcare, podcasts, creators, doesn't matter. Start building some templates for yourself. Then start building for friends for free. Then start asking if they will pay you to build more agents for them to solve more problems for them. You will have competition doing this, but most of them are not using these AI tools yet, so you were able to charge less for as good, if not better service. Your first client is your case study and helps you get your next three clients. After five or ten clients, you can hire a virtual assistant to help you out. After six to 12 months, you scale to 10 to 30 clients and you're making 10 to $30,000 a month. That includes the setup fees and the ongoing maintenance fees. I hope this was helpful. Please check out Hostinger if you found this interesting. They are running a Black Friday sale right now. You will get the lowest price possible. Choose the KVM2 plan. That's the one with enough juice to run multiple AI agents. Deploy N8N, build this content multiplier agent. Then find any creator, any business owner who feels overwhelmed with social media. Implement this into their process. You could set this up for less time than it took to watch this video. This is a very real and realistic opportunity. Most people are just going to watch this and do nothing. Don't be most people. Go do something. Thanks for hanging out on the Kerner office.