GoTranscript
>
All Services
>

Public/how To Set Up An Openclaw Telegram Bot Safely

How to Set Up an OpenClaw Telegram Bot Safely (Full Transcript)

Learn what OpenClaw is, key security risks, and how to deploy a Telegram chatbot on Hostinger’s managed plan with automations like daily YouTube summaries.
Download Transcript (DOCX)
Speakers
add Add new speaker

[00:00:00] Speaker 1: OpenClaw might be one of the most powerful AI tools available right now, but it's also one of the easiest to misuse if you don't understand what it's actually doing. OpenClaw lets an AI assistant run commands, read files, make decisions, and automate real tasks through the tools and services you connect to it. In other words, it can behave like a true digital assistant or even a digital employee. But giving software that much capability also comes with real risks, especially if you're trying to set everything up and manage it yourself. So in this video, in partnership with Hostinger, we're going to take a look at what makes OpenClaw different from other AI tools, what makes it so powerful, and how you can get started using it in a way that's both practical and approachable. And we'll do that by building a simple OpenClaw chatbot you can interact with through Telegram, without having to deal with a complicated setup, server configurations, or managing infrastructure yourself. So first, what makes OpenClaw different from normal AI tools? Tools like ChatGPT, Claude, or Google Gemini are great at analyzing text, answering questions, helping you brainstorm ideas, or generating content. But in most cases, those tools stop at the answer. OpenClaw is designed not just to generate answers, but to actually take actions. It can run tasks, interact with services, schedule actions, and communicate with you through messaging apps like Telegram or WhatsApp, Signal, or iMessage. So instead of just giving advice, it can actually operate inside your digital environment. For example, a typical AI chatbot might help you draft an email response. But an OpenClaw assistant could go further. It could monitor your inbox, detect meeting requests, check your calendar for availability, schedule the meeting, and send the confirmation. Or as another example, you could ask it to monitor your favorite YouTube channels, summarize new uploads each morning, and send you a short digest of the content through WhatsApp. So OpenClaw isn't just a different AI model. What makes it different is that OpenClaw connects AI to real systems and performs real actions. So that's what makes it incredibly powerful, but it's also why it needs to be handled carefully. Because once you give an AI assistant access to files, applications, or services, you're giving it the ability to interact with your digital life in ways that normal AI tools simply can't. And that brings us to the risks. Unlike most AI tools that simply generate text in a browser window, OpenClaw can be given permission to interact with and given access to things like your file system, your API keys, your cloud services, your messaging accounts, and sometimes even your shell or command line. And if that system is misconfigured or exposed to the internet without proper security, it could potentially leak sensitive data or run commands you didn't intend. There are plenty of stories already of employees playing with OpenClaw and accidentally exposing their work computers to the wider internet. Another risk comes from skills or plugins. OpenClaw allows you to install third-party skills that extend what your assistant can do. But just like browser extensions or mobile apps, those skills may request access to services like GitHub, email accounts, or cloud storage. So installing the wrong skill or granting too many permissions could expose part of your system you didn't intend to share. There's also the risk of running OpenClaw directly on your personal computer. Because if the assistant has access to your machine, it could theoretically read files, interact with applications, or trigger automated workflows that affect or compromise your everyday work environment. Now, none of this means you shouldn't experiment with OpenClaw, but it does mean you should approach it thoughtfully and with the right safeguards. And one of the simplest ways to reduce these risks is to run it in an isolated environment rather than directly on your own computer. Instead of setting all that up yourself, platforms like Hostinger handle the environment for you behind the scenes. That means your assistant runs its own secure, always-on space separate from your personal files and applications. So you still get the benefits of isolation and reliability without having to manage servers or configure security or maintain the system yourself. Now the trade-off is that your assistant won't have direct access to apps running locally on your machine. But because most tools like email, calendars, and web services are all cloud-based anyway, OpenClaw can still interact with them and help automate your workflows. So in this video, we're going to build our OpenClaw chatbot using Hostinger's managed OpenClaw plan. If you want to follow along, point your browser at hostinger.com slash startopenclaw. And here you can choose between a fully managed OpenClaw plan or a more advanced VPS setup. For this video, we'll use the managed plan so everything is handled for us and we don't have to worry about any technical setups. You can see some of the advantages here, such as it being ready out of the box, zero maintenance, built-in telegram and whatsapp pairing, and so on. Just click choose plan to go to the cart. Be sure to keep the 24-month period selected for the best rate. Now below that is where you can purchase AI credits. Typically when you run tools like OpenClaw, you have to create accounts with AI providers like OpenAI or Anthropic, generate API keys, and connect those services to your installation. Hostinger simplifies this by bundling AI usage into its own credit system. So instead of configuring multiple AI accounts, you simply purchase Nexos AI credits directly through your Hostinger dashboard. And OpenClaw uses those credits whenever it sends prompts to the AI model. So this means you don't have to configure API keys just to get started. Now you still always have the ability to connect to external AI providers, but the built-in credit system makes it much easier to get started. So make sure you keep this option selected if you're using the built-in system, since you'll need AI credits for your assistant to generate responses. You can also come over here to the right and click have a coupon code, and enter start OpenClaw for an additional discount. And after completing your purchase, you'll see a setup screen while your OpenClaw installation is configured. Once that's done, you'll see the option to connect your WhatsApp or Telegram account to OpenClaw. Connecting to a Discord account will also be an option in the near future. For this example, I'll choose to set it up through Telegram. You'll then see instructions on how to connect your account. And in this case, we'll see the steps to connect to Telegram. Now before we jump into that, you will need your own free Telegram account from which to send and receive messages. If you don't already have one, go to telegram.org to download the app and set up an account. And once you have that, you'll be able to talk to your assistant simply by sending it messages from the Telegram app on your computer or mobile device. And any message you send to your Telegram bot will be routed through OpenClaw and handled by your assistant. So make sure you set up your account, and then we'll come back here and follow these instructions. All right, so I've opened up Telegram on my computer and signed into my account. And here in the chat area, under search, I'm going to search for a user called Botfather. This is Telegram's official tool for creating and managing bots that interact through it. Make sure you do select the one called at Botfather with the blue check mark next to it as some other accounts have similar looking names. Once I've selected that account, I'll click start. And it sends a default message giving you a list of popular commands you can use with it. Notice the first one here is new bot, which lets you create a new bot. And that's what I want to do. So in the chat area down here at the bottom, I'll enter the command slash new bot. And now we just set it up through a conversation with Botfather. It says, all right, a new bot, please choose a name for your bot. Now this is simply the display name you'll see in your chat list. So you can choose anything you like. Since I'm making an open claw bot, I'll call mine pinchy. Next, Telegram asks for a username for the bot. Now this one has to be unique, and it also has to end with the word bot. So you might have to try several names before you find one that isn't already taken. For example, if I just try to call this pinchy bot, it tells me the name is already taken. So I'll probably have to try a couple of different ones. I'll try pinchy opclawbot. And that one is available. So once that's done, you'll get this confirmation message, and Botfather creates the bot and provides a bot token. This token is essentially the credential that allows OpenClaw to connect to the bot. And notice it tells you to keep your token secure and to store it safely because it can be used by anyone to control your bot if they have this token. So in this case, I'm just going to copy that token. And then I'll come back to my browser, to the hosting or setup page, and paste that where it says enter bot token. I'll click next. Then it tells me to go to my bot and paste in slash start. So I'll come back here and type in slash start. And oops, actually what I was supposed to do was go to the bot itself. So I actually pasted that into Botfather, so it gave me this list of instructions again. But in this case, I need to go to my bot. And I can see the link for that bot right here. And here I'll click start. So you can see now I'm talking to my actual bot that was created. And that gives me this pairing token that I can copy and then go back to my OpenClaw setup and paste in. I'll click approve. And now we see this message that we're all set. Now here we see our access key. I'll click access key to copy to my computer's clipboard. Now you might want to paste this access key into a text file or some other storage area on your computer so you always have access to it. And now I'll click OpenClaw. Now you may be prompted to enter your access key, in which case you can just paste it in since you just copied it. But in this case, I'm now looking at my OpenClaw dashboard. Notice the chat area is selected by default. So I can immediately start chatting by typing in the prompt area down below. I'll just type in hello. Now the message you get back may vary, but it's basically telling me that it's here and responding to my messages. It's asking what are we working on today or do I just want to test that the chat is working. I could say I'm testing that the chat is working. And again, it gives me a confirmation. I could ask it other things like, what is your name? Now in some cases, it might ask you to tell it its name, but I'll ask it, do you know your name? It says I haven't set my name in this workspace yet. What name do you want me to go by? I'll say your name is Pinchy. My name is Garrick. And there it is basically confirming that it knows its name now. Now in other OpenClaw setups, this is a point where you would have to configure your communication app like Telegram or WhatsApp, but we've already taken care of it. And we can test this by going back to Telegram and sending the bottom message of, what is your name? And it confirms that it knows the name I've given it. So you can see that Hostinger made it pretty easy to set up my OpenClaw chatbot through Telegram and I can now continue my conversations here or on my phone instead of through the chat area when logged into my web browser. Now that the Telegram chatbot is working, the real value comes from what you can actually do with it. Right after setup, you can instantly use your assistant for things like research, summarizing information, brainstorming ideas, or drafting messages. For example, right off the bat, I can ask it something like, search some of the most reliable and trusted tech news websites and summarize some of the most important AI headlines in the past 24 hours. And I get a message that it's pulling today's AI headlines from multiple sources. And notice it's not only providing the news, but also it gives you insights on why these stories matter. And I could, of course, ask follow-up questions or have it research additional information about any of these topics. And this works right out of the box with no additional APIs or setup required for your bot to scrape content from the web. Now depending on what and how you ask it, it may ask follow-up questions so it fully understands what you're requesting. But in this case, it understood it pretty well and gave me exactly what I asked for. Now I could even ask it to do something even more involved, like have it monitor my favorite YouTube channel for new content with a prompt like, monitor the YouTube channel at Kevin Stratford and send me a summary of any new videos published in the last 24 hours. Send me a summary every morning at 8 a.m. Eastern time. Now this would be a pretty complex task to set up with a regular AI model, but OpenCloud will understand what I'm asking and it will most likely ask follow-up questions to help it refine the workflow as it builds it in the background. Let's see what happens. Now anytime you see the phrase typing there at the top of the screen, that tells you that your command made it through and it's now thinking and responding. Now I've tried this in the past, it's telling me that it's trying a different approach than it tried last time and letting me know exactly what it's doing. In this case, it's using the RSS feed it was able to get from the channel ID. It was able to see the uploads with the published timestamps, and it gives me a summary of what it's set up. The schedule is every day at 8 a.m., the source is Kevin Stratford's YouTube feed, videos published since yesterday at 8 a.m. Eastern time, and that the videos should be summarized and sent to this chat. Notice it's even offering to run a test run now so I don't have to wait until 8 a.m. tomorrow to see the output. So I'll just say yes. Again we can see at the top of the screen that it's thinking. And there it is. It says in the past 24 hours it's published one called AI Agents that screen clients and reply for you. It gives me a link to the video, tells me when it was published, gives me a description, and even offers key takeaways from the video itself. There's the second video that was published, and again it provides all that same information. And until I tell it to stop, this information will now be delivered through Telegram at 8 a.m. every morning without me asking for the information. And because everything runs on my hosting or plan, my assistant is always running in the background ready to send me updates or respond whenever I message it. Now for more complex or specialized tasks, you can also install skills to give OpenClaw additional abilities. If we come back here to the OpenClaw control area, notice there's a skills section over here. Notice this area called built-in skills. Here you'll find skills that are already installed as well as others that can be added. And these skills extend what your assistant can do. For example, the weather skill lets your chatbot retrieve current weather conditions and forecasts. The Trello skill allows you to interact with the Trello project management system. And if you want your assistant to monitor blogs or RSS feeds, you can install the blog watcher skill. Now we're not going to get into skills in this video, but it's good to know that this is where you can expand what your assistant is capable of doing. But even with the default installation, our bot is already powerful and capable enough to handle an enormous variety of tasks and requests. Now again, one thing it can't do in this setup is directly monitor or control applications running on your personal computer. Because OpenClaw is running in its own managed environment, it's separate from your local machine. The trade-off, however, is improved security and reliability. Your assistant is separated from your personal files and applications, it's always online, and it can continue running tasks in the background even when your computer is turned off. So if you've been thinking about giving OpenClaw a try, be sure to visit the link in the description to run your own OpenClaw assistant on Hostinger and see what your bot can do for you. And if you found this helpful, click the like button and consider subscribing to see more videos like this one. And until next time, I'm Garrick, thanks for watching, and we'll see you soon.

ai AI Insights
Arow Summary
The transcript explains what makes OpenClaw different from typical AI chatbots: it can take real actions by running commands, reading files, and integrating with connected tools/services, effectively acting like a digital assistant/employee. It emphasizes security risks—misconfiguration, excessive permissions, third‑party skills, and running on a personal computer—and recommends isolating OpenClaw in a managed environment. The video then walks through setting up an OpenClaw chatbot on Hostinger’s managed OpenClaw plan, using bundled AI credits (no API keys required), pairing it with Telegram via BotFather, and testing basic chat. It demonstrates practical automations like web research for AI headlines and setting a daily 8 a.m. digest that monitors a YouTube channel via RSS and sends summaries to Telegram. Finally, it notes skills can extend capabilities (weather, Trello, blog watcher) while highlighting the managed-hosting trade‑off: less access to local apps but better security, reliability, and always‑on operation.
Arow Title
Building an OpenClaw Telegram Bot on Hostinger (Safely)
Arow Keywords
OpenClaw Remove
AI agent Remove
automation Remove
digital assistant Remove
security risks Remove
permissions Remove
skills/plugins Remove
Hostinger managed plan Remove
Nexos AI credits Remove
Telegram bot Remove
BotFather Remove
bot token Remove
pairing token Remove
RSS feed Remove
YouTube monitoring Remove
daily digest Remove
web research Remove
summarization Remove
always-on hosting Remove
isolation Remove
Arow Key Takeaways
  • OpenClaw differs from standard chatbots by taking actions in connected systems, not just generating text.
  • Power comes with risk: misconfiguration, exposed services, and overbroad permissions can lead to data leaks or unintended command execution.
  • Third-party skills/plugins should be treated like apps/extensions—review permissions and trustworthiness.
  • Running OpenClaw in an isolated managed environment reduces risk versus installing it directly on a personal machine.
  • Hostinger’s managed OpenClaw plan simplifies setup with out-of-the-box configuration and bundled AI credits, avoiding initial API-key management.
  • Telegram pairing requires creating a bot with BotFather, securing the bot token, then completing OpenClaw pairing via a pairing token.
  • After setup, OpenClaw can perform web research/summaries immediately and can schedule recurring automations like daily YouTube digests via RSS.
  • Managed hosting trades local-machine control for improved security, reliability, and always-on background execution.
Arow Sentiments
Neutral: The tone is instructional and pragmatic: it highlights OpenClaw’s power with cautious warnings about security and permissions, then provides a step-by-step setup guide and examples of useful automations without strong emotional language.
Arow Enter your query
{{ secondsToHumanTime(time) }}
Back
Forward
{{ Math.round(speed * 100) / 100 }}x
{{ secondsToHumanTime(duration) }}
close
New speaker
Add speaker
close
Edit speaker
Save changes
close
Share Transcript