From Zero to First AI Assistant in 15 Minutes (OpenClaw)

Check this out. I just got a message on Telegram from my AI assistant. And every morning, it sends me a quick brief like this. It tells me what's on my calendar and anything important that needs my attention. All without me having to ask first. And that right there is what makes this so special. Most AI assistants you see online are really just chat bots. They only do something when you open a tab and start typing. This AI assistant that I created can run on its own on a schedule and message me first because it's always online and it's connected to the tools I use every day like Google Workspace, Slack, and Telegram. My name's Caleb and in the next 15 minutes, I'm going to take you from zero to your first AI assistant set up to work just like mine. You don't need to code. You don't need to be technical. You just need to follow the right steps. And that's why we're going to build this together step by step. And by the end, you'll have a real assistant. One that reaches out to you first, remembers what matters, and helps you take action on important tasks. So, what actually is an AI assistant? Because right now, that phrase gets slapped on everything. Some people call ChatGpt or Claude an assistant, but those are really just chat apps. They're great at answering questions, brainstorming, or writing, but they don't do anything unless you open them and type a prompt first. And then on the other side, you've got automations. Automations can be useful, but they're basically just if this event happens, then do this. There's no real understanding, no memory, and no flexibility. And once you start stacking more complex tasks, automations can get pretty confusing very fast. When you think about having an assistant, you don't want it to feel like AI. You want it to feel like a real human assistant. So that means your assistant should be able to do three core things. First, it needs a brain and memory. The brain comes from an AI model through an API like OpenAI, Enthropic, or Gemini. That's the part that can understand what you mean and make decisions. And the memory is what lets it remember what you tell it, like your preferences, your routine, what you're working on, and what matters to you. So that way, you're not starting from scratch every single time. Second, it should always be running 24/7. That would be awful for a real human assistant, but for AI, we can actually make that work. And later in this video, I'll show you exactly how to set it up so it stays online. And third, it needs tools. Tools are what let it do more than just talk to you. Once it has access to tools you already use, like Telegram, Slack, or Google Workspace, it can actually take action. It can run on a schedule, watch for certain things, and message you when something needs attention without you prompting it first. So, that's what we mean by AI assistant in this video. It's not a chatbot you talk to. It's not a simple automation. We're creating a real assistant that remembers, stays online, and helps you handle tasks even when you're not thinking about it. And that's exactly what we're building. And we're going to build it with something called OpenClaw. If you've never heard of OpenClaw before, here's what it is. Open Claw is a tool that lets you create your own AI assistant and keep it running in one place. You connect it to an AI model for the brain. Then you connect it to the tools you use, and then you can set it up to run tasks on a schedule. That's how you get an AI assistant that can actually show up first instead of just sitting in a chat app waiting for you. And the reason OpenClaw is worth using is because it makes all of that surprisingly easy. OpenClaw is one of the first tools that makes setting up a real AI assistant possible for beginners. So that means OpenClaw hits all of the marks we talked about earlier for what qualifies as a real AI assistant. And the nice part is you can manage all of this from one dashboard, which makes setting things up and staying organized way easier. Now, OpenClaw is an incredibly powerful tool, which means we do need to talk about security for a minute. If you don't treat security seriously with this, someone with bad intentions can easily mess with your assistant, access things you connected, and in the worst case, cost you a lot of money and stress. So, first, let's talk about tokens. In a few minutes, you're going to learn how to connect things like your gateway token, an API key for your AI model, and tokens for connecting tools like Telegram. You want to treat your tokens basically like passwords. If someone gets access to them, they could potentially access your OpenClaw setup, impersonate your assistant, connect to your integrations, and run things you didn't approve. So, we're going to be careful with these tokens, and save them somewhere safe. The second thing to be careful of are skills. Skills are what give your assistant the ability to do things like connect to services, run tasks, or pull information. The most common place to get these skills is from a website called Clawhub or GitHub. But because OpenClaw is still new and growing, there are people out there uploading malicious skills designed to steal info, grab your tokens, or do shady stuff in the background. So, just make sure you only install skills you trust. And always check what a skill is doing before you connect it to anything important. Third, you need to be careful about where you install OpenClaw. You can run OpenClaw on your personal computer, but I don't recommend it. If you install an Always On assistant on the same machine you use every day with your personal accounts, files, and login, you're increasing your security risk in a big way. That's why I recommend using something called a VPS. A VPS is basically a separate computer in the cloud that you rent, and it keeps OpenClaw isolated from your personal device. It also has a second benefit. A VPS lets your assistant run 24/7, which is kind of the whole point. And that's why in this video, we're going to be using a VPS. And for beginners, the easiest way to set up an OpenClaw VPS is with Hostinger. That's because they have a one-click OpenClaw template that deploys everything automatically and gets you set up fast. So, go to the link you see on screen right now or click the first link in the video description. That's going to take you to the correct page for setting this up. It's also going to save you an extra 10% off any VPS plan. And once you go to that link, you'll be brought to this page right here. Start by choosing your VPS plan. By default, this will be set to KVM2. However, if you want to start small, you can also set this to KVM1 and upgrade your plan later on if you need more resources. Once you choose your plan, go down and click on deploy. On this page here, start by choosing your billing period. By default, this will be set to 24 months, and that's going to maximize your savings. However, you can also switch this from 12 months and 1 month plans as well. I wouldn't recommend choosing the one-mon plan because it ends up being way more expensive later on. So, to stay in the middle and still get plenty of savings, I'm going to go with the 12-month plan. Choosing that plan will also still get us access to the extra discount. Scrolling down, you'll see this option here that says ready to use AI. I'd recommend turning this off. We're going to add our own API keys later on in the video. You'll also see an option for daily auto backups. This is a great option if you want to add it, but I'm going to teach you how to create manual snapshots within your hosting or dashboard, so that way you can manually back up your VPS. Lastly, choose your server location. Just choose the option that's closest to you. Once you're done here, scroll back up to the top and click on continue. From there, you'll create an account with Hostinger and go through the payment process. Once you're done with payment, you should land on the configuration page for your OpenClaw deployment. On this screen, you'll see a field that says OpenClaw gateway token. Go ahead and copy that. Make sure you keep this safe and don't share it with anyone. Treat it like a password. If someone has your gateway token, they can access your OpenClaw dashboard, which would not be good. So, copy it and store it somewhere safe. Next, we need to give OpenClaw a brain. This is where your AI model comes in through an API key like OpenAI, Enthropic, or Gemini. On the same hostinger page, you should see a field where you can paste in your API key. If you already have an API key, paste it in here. If you don't, you'll need to go to the AI model provider site, create a key, and add credits or billing so it can actually run. I'm going to be using Enthropic for my Open Claw Assistant. I'll also leave links in the description for the other API key dashboards just in case you're not sure where to go to create this. One quick thing to understand here is that using an API costs money because every request you send to the AI model is build based on usage. So, with your model's billing settings, make sure you set a spend limit. You don't want to accidentally leave this unlimited and end up with a huge surprise bill. So, set a limit you're comfortable with, save it, and then come back to Hostinger. All right. Now that your API key is pasted in and you've got your spend limit set, the last step on this page is to deploy. Go ahead and click on deploy, and Hostinger is going to deploy OpenClaw automatically using the template. This can take a few minutes to complete, so I'm going to skip ahead once it's finished. Awesome. As you can see, our OpenClaw container is now running. What you're looking at here is Hostinger's Docker Manager. In simple terms, this is the control panel for anything you run using Docker. Right now, we're using it for OpenClaw, but later on you can also deploy other apps the same way. If you ever need to manage your OpenClaw setup or find your tokens again, click manage. Then scroll down to the environment tab and expand it. This is where you can view your OpenClaw gateway token again along with any API keys and other settings. If you ever need to add or update environment variables in the future, this is where you'll do it. Next, to open your OpenClaw dashboard, scroll back up the page and click on this purple link here. Paste in your OpenClaw gateway token. Then click login. Once you're in, you'll land on the OpenCloud dashboard. On the left, click into overview and make sure your status shows as connected. If it doesn't, paste in your gateway token here and click connect. Now, let's connect OpenClaw to Telegram. So, instead of using the dashboard, you can message your assistant straight from your phone. First, find the OpenClaw chat window on the lefth hand side. Then, type in let's set up Telegram. OpenClaw is then going to respond with the exact next steps you need to follow. It says here we need to start a conversation with BotFather on Telegram. I'm going to do this part using the Telegram desktop app, but once we're done, you'll also be able to message your assistant from the mobile app as well. So, let's open Telegram and search for atbotfather. Botfather is what lets you create and manage your own bots, which is super important for setting this up with OpenClaw. Start a chat with BotFather, then click the new bot command. It'll ask you for a bot name and username. Pick whatever you want here. Once it's created, BotFather is going to give you a bot token. Go ahead and copy that bot token and just like the gateway token, treat this like a password. Don't share it. Don't post it anywhere and store it somewhere safe. Now jump back over to OpenClaw. Paste in that bot token here and send it. Next, we'll need to pair your Telegram account with your OpenClaw setup. So head back to Telegram and click the link from BotFather to open your new bot. Then click start. Your bot should send you a message with your Telegram user ID and a pairing code. Go ahead and copy both of those lines and go back to OpenClaw. Paste that information into the chat here and send it. Okay, now at this point, everything should be configured properly. Let's go ahead and test it to make sure it's working. Send a message to your bot in Telegram and see if it responds. If everything is connected the right way, you should see your assistant reply back in Telegram. And once you see that response, that means you're officially connected. Next, let's learn how to install skills. Skills are what turn openclaw from something that can just chat into something that can actually do tasks using real tools. And the nice part is you don't have to figure out how to install skills all on your own because OpenClaw can actually help you install skills and troubleshoot them. So I'm going to do this through Telegram. First I'm going to type what skills can I install. OpenClaw will usually point you towards the skills library. It can also tell you what it recommends based on what you're trying to do. In our case, we're building a daily brief that can read from Google Workspace, like your calendar and your emails. So, the main skill we want is the Google Workspace skill called GOG. So, next, I'm going to ask, can you help me install the GOG skill and set up Google Workspace Access? From here, the assistant will walk you through the steps for that specific skill. And this is important to understand. Every skill is a little different. Some are one-click installs, some need environment variables, some need permissions, and some require a quick authentication flow. So, if anything doesn't work right away, don't panic. Just work with the assistant and follow the steps it gives you. If you run into an error, copy the error message into the chat and ask it what to do next. OpenClaw can actually rewrite parts of its own skill code and help troubleshoot skill installations, so you can usually fix issues pretty easily. So now I'm going to follow the install steps from my assistant and connect Google Workspace so that way the assistant can access my calendar in Gmail. All right, now that I've installed the Google Workspace skill and given OpenClaw access to my Gmail and calendar, we're going to create a job that makes this feel like a real assistant. Creating jobs is what tells OpenClaw to run on its own on a schedule and message you first without you having to prompt it. So back in our Telegram chat with OpenClaw, I'm going to ask it to create a job for us. I'm going to type, "Help me create a daily brief job. Every morning at 9:00 a.m., I want you to check my Google calendar for the day. Scan my Gmail inbox for anything important. Then send me a short summary through Telegram with my top priorities." From here, the assistant will walk us through setting everything up. Once that's set up, we're going to do one quick test just so you can see what this brief is going to look like every morning. So, I'm going to ask it to run the daily brief right now. And there we go. My assistant has now performed the job perfectly. It scanned through my Google calendar and found my team meeting for the week. Then it looked through my Gmail inbox and found an email from John. And in that email, it says he can't make the team meeting this week. Then it provided me my list of priorities, which is to review John's email and adjust the meeting time. And now going forward, my assistant will perform this job every morning at 9:00 a.m. all on its own. And now using this exact same method that I just showed you, you can install skills and create jobs and solve all sorts of problems within your day-to-day life. Okay, now that we've created our first job, let's run through a quick final security checklist. These are a few extra steps you can do right now to make your setup safer and make sure you're prepared if anything ever goes wrong. First, go to OpenClaw's official security page, which I'll leave linked in the video description. Copy that link, paste it into your OpenClaw chat, and ask your assistant to implement and verify the recommended security settings. The reason you do this is so OpenClaw can secure itself using the official best practices. You can quickly confirm everything is set correctly. Next, set approval rules for risky actions. Tell OpenClaw that if it's going to send a message on your behalf, it must draft it first and ask for your approval, and that it should always ask before deleting files or making network requests. The reason you do this is so it can't take destructive actions without your permission. Next, add cost guard rails inside of OpenClaw. Set a retry limit so if a task fails a few times, it stops and set a runtime limit so tasks can't run forever. The reason you do this is to prevent expensive loops that can rack up API charges. Also, make sure you take a manual snapshot in your hosting or dashboard before you start experimenting with new skills or major changes. You want to do this because if something breaks, you can restore it back to a working version in minutes instead of rebuilding everything from scratch. And speaking of Hostinger, if you ever need to manage your OpenClaw setup, that's where you're going to do it. That's where you can find your VPS settings, Docker Manager, environment variables, backups, and security settings. So, if you ever need to update a token, swap your AI model key, restart your OpenC container, or roll back to a previous version, you'll do it all right here inside of Hostinger. And also, if you're following along and you haven't set up your VPS yet, make sure you use the first link in the video description. That link takes you to the correct hosting or page for this OpenClaw VPS setup, and it will give you the most affordable deal available. If you run into any issues or you have questions about your setup, just leave a comment down below, and I'll try to help you out. Thanks for watching this video and I'll see you in the next